Geographical diversification and Solvency II A proposal by Lloyd’s The proposal This is a proposal for the recognition of geographical diversification in the non-life underwriting risk module of the SCR standard formula. Why should the standard formula recognise geographical diversification? Recognition of geographical diversification reflects Solvency II’s intentions because: Solvency II is based on an economic approach to the setting of capital, an approach that requires recognition of diversification. A key form of diversification for insurance undertakings is diversification of their underwriting portfolios by location – geographical diversification. This recognises reduced correlation between losses in different places – e.g. losses arising from bad weather in the UK have no connection to claims experiences in Greece. The Solvency II Directive therefore makes explicit provision for diversification. Insurance undertakings carrying on business in different regions across the world lessen the concentration of risk that occurs if they are restricted to a single country. This has a material impact on the amount of capital they require. A detailed rationale for recognition of geographical diversification is set out in Appendix 1. Recognition of geographical diversification by Solvency II The non-life module in QIS4 included geographical diversification. CEA’s January 2009 QIS4 feedback to the European Commission said: “The introduction of geographical diversification into QIS4 was a valuable improvement in the SCR standard formula compared to QIS3 - it enabled a much more realistic recognition of insurers' risk profiles.” However, subsequently, CEIOPS decided not to include geographical diversification. Its advice on non-life underwriting risk1 said: “While this change is crucial for reinsurers and cross-border groups, it was seen as introducing unnecessary complexity at solo level, in view of the materiality of the reduction in capital requirement they could obtain from the calculation.” This decision elicited much comment when it was first outlined in CP48 (see Appendix 2). CEIOPS’ report on QIS4 recognised that geographical diversification is a material factor 1 DOC-41-09, October 2009 1 for many solo entities such as reinsurers and specialists in transport insurance, but has little or no effect on others. CEIOPS recognises that undertakings with geographically diversified portfolios are seriously affected by the non-recognition of geographical diversification. It means that the standard formula is not an appropriate method of calculating their SCR and requires them to use undertaking-specific parameters (if appropriate) or to develop internal models. On the other hand, the issue is less relevant to undertakings carrying on business on a regionally restricted basis, who want to avoid additional complexity in their SCR calculations. The EU’s insurance sector includes many undertakings which carry on insurance business across the world: important and successful components of the EU’s financial services industry. The standard formula is intended “…to reflect the risk profile of most insurance and reinsurance undertakings” (Directive preamble 26) and “…to enable all insurance and reinsurance undertakings to assess their economic capital” (Directive preamble 65). The removal of geographical diversification means that the standard formula does not reflect the risk profile of globally active undertakings and cannot be used by them to assess their economic capital. The aim is to incorporate geographical diversification in the standard formula so that it can reflect the risk profiles of globally active undertakings, whilst avoiding the imposition of complexity on undertakings whose business is not geographically diversified. This proposal does this in two ways: It includes a simplification whereby undertakings finding allocating risks to geographical regions too complex can report all risks within a single region. It makes the approach to geographical diversification less complex than that adopted by QIS4. The number of regions into which premiums and reserves must be broken down is reduced from 54 to 18. Geographical diversification: specification The approach set out here is similar to, but not exactly the same as, that set out in the QIS4 specification. Incorporation of geographical diversification into the non-life premium and reserve risk module would be standard with an acceptable simplification of allocating all risk to one region, at the option of the undertaking. If an undertaking wishes to claim geographical diversification benefits it will calculate a Herfindahl index based on the geographical location of the risks underlying the premiums and reserves for each line of business (LoB). The Herfindahl index (also known as the Herfindahl-Hirschman Index) is commonly used to calculate the size of firms in relation to an industry and so to indicate the extent of competition within the industry. Here it is used to calculate the geographical concentration of a portfolio of premiums and reserves. The maximum value of the Herfindahl index is 1, implying no 2 geographical diversification. The closer the Herfindahl index is to 0, the less concentrated the portfolio is and the greater the diversification effect. The index which represents a measure of the implicit geographical diversification is calculated for each line of business as follows: (V ( prem , j ,lob ) DIVpr ,lob V( res , j ,lob ) ) 2 j (V( prem , j ,lob ) V( res , j ,lob ) ) j 2 In this formula, for each individual line of business, V(prem, j, lob) and V(res, j, lob) are the standard non-life volume measures for premium and reserve risk, respectively split further by the underlying geographical regions which are represented by j. The overall volume measure V to be used in the standard formula calculation would then be determined as follows: V V Lob lob Vlob is the line of business volume measure for premium and reserve risk which can be defined as: Vlob = ( V(prem,lob) + V(res, lob)) * (0.75+0.25 * DIVpr,lob) For undertakings choosing not to claim geographical diversification, a simplification would be applied whereby all data was entered within one geographical region and DIVpr,lob would implicitly be set to 1 for all lines of business. Maximum diversification benefit The maximum diversification benefit available would be 25%. Appendices 3 – 5 discuss the factors underlying the selection of this figure. This does implicitly assume moderate (50%) correlation between non-identical regions. Materiality threshold If an undertaking has more than 95% of its non-life activities (premium and reserves) in the same geographical area, it cannot benefit from geographical diversification. Lines of business Geographical diversification can be applied to any line of business, other than Credit and suretyship. 3 Non-proportional reinsurance classes will have less need for explicit recognition of geographical diversification as this is better reflected in the calibration of the standard factors. This leaves two options for non-proportional reinsurance: To reduce the maximum allowance from 25%; or To remove geographical diversification entirely. Lloyd’s believes some additional diversification would remain and proposes a maximum of 10% to recognise this. Regional breakdown We consider that the 54 regions contained in QIS4 can be streamlined to make calculation easier. Our approach is based on the United Nations geo-scheme, developed by the UN Statistics Division for statistical purposes, which divides the world into “macrogeographical regions”. The UN confirms that these geographical divisions do not imply any assumptions regarding political or other affiliations of countries or territories. The geo-scheme consists of 22 sub-regions. From an insurance point of view it is possible to aggregate these further, to take account of the size of national markets. In addition, we believe that it is necessary to take special account of the US, in view of the size of the US non-life market and of South America, in view of its importance to some geographically diversified EU insurers. The US can be divided into regions using the NAIC’s regional split. The regional split that Lloyd’s recommends is as follows: 1. Central & Western Asia (UN geo-scheme Central Asia and Western Asia, less Cyprus) Armenia Azerbaijan Iraq Israel Kuwait Kyrgyzstan Palestinian Territories Qatar Tajikistan Turkey Uzbekistan Yemen Bahrain Jordan Lebanon Saudi Arabia Turkmenistan Georgia Kazakhstan Oman Syrian Arab Republic United Arab Emirates 2. Eastern Asia (UN geo-scheme Eastern Asia) China Mongolia Hong Kong North Korea Japan South Korea 4 Macao Taiwan 3. South and South-Eastern Asia (UN geo-scheme Southern Asia and SouthEastern Asia) Afghanistan Cambodia Lao PDR Nepal Sri Lanka Bangladesh India Malaysia Pakistan Thailand Bhutan Indonesia Maldives Philippines Timor-Leste Brunei Darussalam Iran Myanmar Singapore Vietnam 4. Oceania (UN geo-scheme Oceania region) American Samoa Australia French Polynesia Guam Micronesia Nauru Niue Norfolk Island Papua New Guinea Pitcairn Tokelau Tonga Wallis & Futuna Islands Cook Islands Kiribati New Caledonia N. Mariana Islands Samoa Tuvalu Fiji Marshall Islands New Zealand Palau Solomon Islands Vanuatu 5. Northern Africa (UN geo-scheme Northern Africa and Western Africa plus Cameroon, Central African Republic and Chad) Algeria Cape Verde Egypt Guinea-Bissau Mauritania Saint Helena Togo Benin Central African Rep. Gambia Liberia Morocco Senegal Tunisia Burkina Faso Chad Ghana Libya Niger Sierra Leone Western Sahara Cameroon Cote d’Ivoire Guinea Mali Nigeria Sudan 6. Southern Africa (UN geo-scheme Southern Africa, Eastern Africa and Middle Africa other than countries specified under Northern Africa) Angola Dem Rep of Congo Ethiopia Madagascar Mozambique Rwanda South Africa Tanzania Botswana Djibouti Gabon Malawi Namibia Sao Tome & Principe Swaziland Zambia Burundi Equatorial Guinea Kenya Mauritius Rep of the Congo Seychelles Uganda Zimbabwe 5 Comoros Eritrea Lesotho Mayotte Reunion Somalia United Rep. of 7. Eastern Europe (UN geo-scheme Eastern Europe) Belarus Moldova Slovakia Bulgaria Poland Ukraine Czech Republic Romania Hungary Russian Federation 8. Northern Europe (UN geo-scheme Northern Europe) Aland Islands Faeroe Islands Republic of Ireland Lithuania United Kingdom Channel Islands Finland Isle of Man Norway Denmark Guernsey Jersey Svalbard, Jan Mayen Estonia Iceland Latvia Sweden 9. Southern Europe (UN geo-scheme Southern Europe, plus Cyprus) Albania Cyprus Macedonia San Marino Vatican City Andorra Gibraltar Malta Serbia Bosnia Greece Montenegro Slovenia Croatia Italy Portugal Spain 10. Western Europe (UN geo-scheme Western Europe) Austria Liechtenstein Switzerland Belgium Luxembourg France Monaco Germany Netherlands 11. Northern America excluding the USA (UN geo-scheme Northern America, less the USA) Bermuda Canada Greenland St Pierre & Miquelon 12. Caribbean & Central America (UN geo-scheme Caribbean and Central America) Anguilla Barbados Costa Rica El Salvador Haiti Mexico Panama St Lucia Turks & Caicos Is’ds Antigua & Barbuda Belize Cuba Grenada Honduras Montserrat Puerto Rico St Martin US Virgin Islands Aruba British Virgin Islands Dominica Guadeloupe Jamaica Netherlands Antilles St-Barthelemy St Vincent 6 Bahamas Cayman Islands Dominican Republic Guatemala Martinique Nicaragua St Kitts & Nevis Trinidad & Tobago 13. Eastern South America (UN geo-scheme South America divided) Brazil Paraguay Falkland Islands Suriname French Guiana Uruguay Guyana 14 Northern, southern and western South America (UN geo-scheme South America divided) Argentina Ecuador Bolivia Peru Chile Venezuela Colombia 15. North-east US (NAIC North-eastern zone) Connecticut Maryland New York Delaware Massachusetts Pennsylvania District of Columbia New Hampshire Rhode Island Maine New Jersey Vermont 16. South-east US (NAIC South-eastern zone, less US Virgin Islands) Alabama Kentucky Puerto Rico W. Virginia Arkansas Louisiana South Carolina Florida Mississippi Tennessee Georgia North Carolina Virginia Iowa Missouri Oklahoma Kansas Nebraska South Dakota 17. Mid-west US (NAIC Midwestern zone) Illinois Michigan North Dakota Wisconsin Indiana Minnesota Ohio 18. Western US (NAIC Western zone, less American Samoa and Guam) Alaska Hawaii New Mexico Washington Arizona Idaho Oregon Wyoming California Montana Texas Colorado Nevada Utah The suggested regional split is indicated on the attached maps (Appendix 6). Regional non-life insurance markets’ relative sizes are indicated below. This shows the regions more likely to be significant sources of business to EU insurers and reinsurers transacting business globally. Regions are not intended to be of broadly equal size, but to be sufficiently large to justify their separate treatment. The proposal that countries be grouped into regions is an integral part of this proposal. Nevertheless, we are receptive to suggestions, from supervisors, insurers and other 7 interested parties, for change to the precise number of regions and their make up, if alternatives are considered to be more satisfactory. 2008 Non-life premium income (Source: Sigma 3/2009) USD bn Central & West Asia Eastern Asia South & South –eastern Asia Oceania Northern Africa Southern Africa Eastern Europe Northern Europe Southern Europe Western Europe Northern America (excl. USA) Caribbean & Central America E South America NSW South America North-east US South-east US Mid-west US Western US 25.3 192.5 28.0 34.6 6.2 10.5 69.4 157.4 115.5 355.7 57.7 14.1 27.4 24.4 165.3 165.3 165.3 165.3 Note: figures for US are approximations only, derived by dividing total US non-life premiums of USD 661.2bn between the four regions. Allocation of transactions to particular countries We suggest that the method of defining risk location should be: Simple and easy to apply. Objective and not open to subjective interpretation. At the overall level, give a good approximation of the spread of risk in the undertaking’s account. We therefore suggest the following methods of allocation: Motor – location of insured vehicle registration. Fire & other property damage – physical location of the insured risk. General liability – insured’s location, unless the coverage provided is: 8 (i) Products liability or professional indemnity; and (ii) The insured provides goods or services to regions other than its home region and (iii) The main risk of loss arises in a region other than its home region. The risk should then be allocated to the region where the greatest risk of insured loss arises. All other classes (including MAT) – determined by the undertaking to reflect the location of loss. Depending on the precise way the contract is written, this may be the insured’s location or the location at which an insured asset has its greatest exposure to insured perils. It is important to bear in mind that geographical diversification operates at an aggregate portfolio level, rather than at the level of individual risks. Some of the factors that give rise to geographical diversification are a consequence of diversity in the physical locations at which losses arise; others, such as diversity in risk and loss cultures and legal environments reflect the countries in which insurance undertakings are carrying on business. Any rule will produce anomalies, but these will not give rise to serious distortions at the aggregate portfolio level. The premium and reserves for a single contract insuring risks located in several regions (“global contracts”) should be split and allocated to regions on the basis of the underlying locations of risks. A contract insuring £100m of risk spread across locations in 10 countries worldwide evidences reduced risk correlation, in comparison with a contract insuring £100m of risk in a single or small number of locations in one country/region. Lloyd’s April 2010 9 Appendix 1 Rationale for recognising geographical diversification Diversification and insurance Solvency II is based on an economic approach to the setting of capital, an approach that requires recognition of diversification. The European Commission’s Impact Assessment Report2 says (page 104): “An economic risk based approach also takes account of the specific risk profile of the insurance undertaking and the impact of risk mitigation techniques, as well as size and diversification effects.” This reflects international thinking on insurer capital requirements. The IAIS Common Structure for the Assessment of Insurer Solvency (February 2007) says (para 79): “The IAIS would thus suggest that an overall capital requirement should take into account diversification between risk factors where this can be substantiated with sufficient robustness.” The Chief Risk Officer Forum (CRO) report “A framework for incorporating diversification in the solvency assessment of insurers” (June 2005) points out that concentration of risk is bad for the insurance industry and consumers and that diversifying strategies are the basis of sound risk management. It further states that existing regulatory solvency approaches do not adequately take diversification into account, an issue which it considered regulators should address. Diversification and the Solvency II Directive The Solvency II Directive makes explicit provision for diversification. Directive Article 13(39) defines “diversification effects” as: “…the reduction in the risk exposure of insurance and reinsurance undertakings and groups related to the diversification of their business, resulting from the fact that the adverse outcome from one risk can be offset by a more favourable outcome from another risk, where those risks are not fully correlated” Preamble 64 says: “…economic capital should be calculated on the basis of the true risk profile of those undertakings, taking account of the impact of possible risk-mitigation techniques, as well as diversification effects.” Article 104(4) includes the following: 2 [SEC 2007] 871 10 “Where appropriate, diversification effects shall be taken into account in the design of each risk module.” Geographical diversification An important form of diversification for insurers is diversification across locations, i.e. geographical diversification. A geographically diversified portfolio shows lower correlation of loss than one which is limited to a single country. Insured losses arise in particular locations at particular times and for the majority of classes the underlying sources are not dependent. Loss frequency and severity reflect factors that vary geographically, such as catastrophe exposure, physical geography, climate, culture, legal systems and levels of prosperity. The KPMG study of insurance prudential supervision, prepared for the European Commission in May 20023, commented: “International companies experience risk reduction due to geographical diversification of risks.” This study discussed at some length the diversification of a portfolio of insurance risks, including geographical diversification. The CRO June 2005 report classified diversification benefits into four distinct categories, including “Level 4 – Across geographies or regulatory jurisdictions” (section 4.4). Its survey of reports analysing the historical impairments of insurance companies concluded: “…these spikes in insolvencies were largely driven by insurers with highly concentrated risk profiles (either geographically or in terms of asset mix), which were particularly adversely impacted by the large loss events of these periods. Geographical concentration is particularly important. For example, Hurricane Andrew led to higher impairment rates among P&C insurers, including 11 insolvencies. Yet no large multi-state insurer (e.g. more geographically diverse) was impaired, even though many such insurers did suffer losses from the same natural catastrophe event.” The European Central Bank published a report entitled “Potential impact of Solvency II on financial stability” in July 2007. This assumed that Solvency II would recognise risk diversification across locations and saw this as contributing to a positive expected outcome of improved efficiency and competitiveness of EU insurers4. It contrasted this with existing insurance regulatory systems in EU countries which: “…do not adequately account for risk diversification in the insurance business, so that the risk of insurance groups being engaged in many different…geographical areas could potentially be overestimated, and capital requirements may appear artificially high.” 3 4 Contract no: ETD/2000/BS-3001/C/45. See para 3.4.16 Section 3.1 11 Appendix 2 Reactions to the exclusion of geographical diversification from the Standard Formula NonLife underwriting module The following are responses received by CEIOPS to its decision in CP 48 “Standard Formula – Non-life underwriting risk” not to apply geographical diversification. They are taken from CEIOPS website. Lloyd’s own comments are not included. Organisation Comment Association of British Insurers (ABI) We strongly disagree with CEIOPS. There is a strong case for recognising geographical diversification. Omitting recognition would be a serious departure from the Directive and lead to substantial additional prudence. AMICE We suggest recognizing geographical diversification as was done during QIS 4. Geographical diversification should be recognised using a blending formula for business underwritten or commitments existing in different geographical areas. Belgian Coordination Group Solvency II (Assuralia/RABA) No allowance for geographical diversification for non-life business will be applied. This will decrease as well the incentive to spread risks over different geographies. CEIOPS should review the decision to exclude geographical diversification. CEA All diversification effects need to be considered appropriately in the standard formula. The introduction of geographical diversification in QIS4 was a valuable improvement to the SCR formula. Failing to recognise it would decrease the incentive undertakings have to spread risks across different geographies. Ceiops should review the decision to exclude geographical diversification and should reward sound risk management. CRO Forum Danish Insurance Association Not allowing for geographic diversification is not recognising one of the key principles of insurance, which is giving credits for well diversified portfolios. The decision to not apply geographical diversification for non-life business across the globe as proposed in the QIS 4 exercise should be reconsidered. We agree that some geographical diversification benefits will be reflected in the estimated volatilities when calibrated on European historical data, however, excluding the possibility for company specific geographical diversification is not in line the level 1 directive and the economic risk-based approach to solvency. FFSA FFSA thinks that geographical diversification should be taken into account. GDV Diversification effects should be considered appropriately in the standard formula. There is a strong case for recognising geographical diversification. 12 Groupama We suggest recognizing geographical diversification as it was done during QIS 4. Being geographically well-diversified is an important element which reduces risk exposure Groupe Consultatif …although removal of the diversification benefit might well simplify the application of the SCR but, at the same time, it could increase the SCR for some insurers by amounts that might be material. Do not agree that this should be left to internal models. Small companies and captives may write significant amounts of business across borders (e.g. in Ireland many companies write north & south of the border). Not allowing for geographical diversification will push up their capital requirement and put them at a disadvantage. International Underwriters’ Association (IUA) We are disappointed that geographical diversification is not adequately allowed for in this module…We believe that it is important for the economic realities of risks written are recognised in the standard formula, especially if some companies rely on the standard formula whilst gaining internal model approval. ..We do not believe the complexity issue is a relevant one; KPMG We disagree with the proposal to exclude geographical diversification as it: a. goes against theory b. goes against the principles of Solvency II (an economic assessment) c. goes against the views of various respected international associations such as the IAIS and IAA d. actively discriminates against a significant portion of the EU insurance/reinsurance market. That is the large, cross border or reinsurance undertakings e. implies that certain firms will get internal (or partial internal) model approval or will use undertaking specific parameters. This is an inappropriate assumption when forming the standard formula parameters and approaches f. incorrectly states the alternatives are complex or impractical. There are alternative that are completely aligned with the principle of proportionality (in that only those who will benefit have to do any significant extra work) g. proposes implicitly allowances that will be inadequate if not calibrated correctly… h. proposes implicit allowances that will knowingly (and avoidably) understate the capital requirements for a large number of (re)insurance undertakings i. ignores realistic improvements to the QIS4 approach (rather than the alternative suggested). The introduction of geographical diversification in QIS4 was widely welcomed Legal & General We strongly disagree. We believe that geographical diversification should be allowed, and is under the level 1 directive Munich Re Geographical diversification should be recognised also in the standard formula as this is one of the main principles of insurance. Companies should not be forced to use partial internal model in 13 order to recognise this positive feature of a balanced portfolio. RBS Insurance Whilst not currently material for us, we believe that allowing geographical diversion for catastrophe risk is important, and in keeping with solvency II principles. We believe policyholders will be better protected when their insurers (and in particular the catastrophe reinsurers of the insurers) have a geographically diversified portfolio. ROAM We suggest recognizing geographical diversification as it was done during QIS 4. Geographical diversification should be recognised using a blending formula for business underwritten or commitments existing in different geographical areas. An important argument is an argument of level playing field. Recognizing geographical diversification on solo level is necessary to allow companies with foreign branches to be treated on diversification level equivalent to companies with subsidiaries who file group SCRs etc... RSA Insurance Group We do not agree with the conclusion to drop geographical diversification. UNESPA- Association of Spanish Insurers and Reinsurers Diversification effects should be considered appropriately in the standard formula There is a strong case for recognising geographical diversification. The introduction of geographical diversification into QIS4 was a valuable improvement to the SCR formula. Omitting recognition would be a serious departure from the Directive and lead to substantial additional prudence. XL Capital Group The removal of geographical diversification is very disappointing and goes against the interests of firms writing large volumes of global risks and global programs. We do not believe that “complexity” is a valid ground for its exclusion, particularly given the level of complexity noted elsewhere in the standards formula. 14 Appendix 3 Allowance for geographical diversification A derivation of optimal geographical diversification allowances to apply under Solvency II is covered in an ASTIN paper by Hürlimann (2009)5. We recommend interested parties refer to this paper. Lloyd's has deduced two key findings from the papers that are relevant to the current discussion: CEIOPS’ 25% maximum diversification used in QIS4 is equivalent to assuming a 50% correlation between regions. A rework of this derivation is provided in Appendix 4 25% is not optimal and using a range of plausible underlying distribution assumptions the maximum credit should be between 30-33%. Lloyd's has supplemented the theoretical results with two illustrative examples to show the extent that diversification does exist within lines of business. An outline of the example is: - conducted for two major lines of business (Property & Casualty) - based on largest 5 currencies to give approximate geographical diversification without making any assumptions on exact geographical splits - only designed to support that the proposed levels are appropriate - based on 1:200 year reserving risk to estimate effects of diversification within a given line of business This analysis shows similar results for both property and casualty lines which are consistent with theory. Specifically that, even using a simplified approach, the implicit diversification is 33% for casualty and 32% for property. This is consistent with the theoretical findings of Hürlimann. Further details of the analysis are shown in Appendix 5. Lloyd's proposes the maximum allowance of 25% for geographical diversification (as used in QIS4) remains. The rationale for this is: 5 - It is a prudent approach. - The underlying assumption of 50% correlations between regions appears reasonable. - A theoretical derivation suggests a more appropriate maximum of 33%. - Lloyd’s illustrative example supports the theory. http://www.actuaries.org/ASTIN/Colloquia/Helsinki/Papers/S3_3_Hürlimann.pdf 15 Appendix 4 Derivation of maximum allowance for geographical diversification assuming 50% correlation between selected regions Let V n V j 1 j be the geographical decomposition of the volume measure of a line of business into n geographical regions. Let us assume that the correlation coefficient (rho) between any two non-identical regions is 50%. This can be represented by ij 1, i j 1 1 ij , ij , 2 2 0, i j (X.1) and that diversification can be measured by the intra-portfolio correlation coefficient n n Q ij wi w j [1,1], wi i 1 j 1 Vi , V (X.2) where wi represent the portfolio weights of the non-life risks in the geographical regions. If we use H n w i 1 2 i to denote the Herfindahl-Hirschman index, and based on assumption (X.1), Q can be expressed as 1 Q (1 H ) . 2 (X.3) The economic capital of the insurance risk portfolio to the confidence level is supposed to depend only on the random loss L and is denoted by EC [L] . In the standard Solvency II approach, the economic capital is defined to be the value-at-risk ( VaR ) of the random loss taken at the confidence level 99.5% , so we have EC [ L] VaR[ L] . Adjusting for geographical diversification the QIS4 non-life risk capital can be represented as 1 (1 Q) EC [ L] . 2 (X.4) It can bee seen that if Q 1 (perfect positive dependence between the regions), no reduction for diversification occurs while if Q 1 (perfect negative dependence), the nonlife risk capital charge vanishes. And we assume a linear dependence structure between these two perfect dependences. In this simple model, based on (X.3) and (X.4), the diversified non-life risk capital charge reads (0.75 0.25 H ) EC [ L] . 16 (X.5) The maximum diversification benefit of 25% can be achieved when H reaches the lower limit of 0. This is consistent with the QIS4 technical specification on geographical diversification cap. 17 Appendix 5 Lloyd’s examples We give two numerical examples, which compare the proposed 25% diversification benefit cap with Lloyd’s own diversification modelling results. These analyses are gross of reinsurance and are included for illustrative purposes – they are designed to show that the theoretical and practical results are similar and support a maximum credit for diversification of at least 25% Lloyd’s market aggregate data covering 1993 to 2009 years of account is used in our analysis. The data is split into five major currencies. This gives approximate geographical diversification without making any assumptions on exact geographical splits. The classes of business being modelled are Casualty and Property (D&F). A standard Bootstrap approach is used to produce the reserve distribution with the mean of the distribution scaled to our best estimate reserves. Required capital is taken as the difference between the 99.5th percentile reserve and the best estimate reserve. The analysis is carried out on an underwriting year basis, which will include an element of unearned and unincepted exposures at the valuation date. The effects of diversification are estimated by comparing the sum of the capital requirements of the individual currencies with the capital requirement when modelling the class as a whole. Table Y.1 and Table Y.2 summarise the diversification benefit results from our modelling exercises. The five major currencies were: AUD, CAD, EUR, GBP, and USD. Please note that the total best estimate reserves were scaled to 100 for consistency. Table Y.1: Diversification Results on Casualty Best Estimate Reserve AUD 7.9 CAD 3.7 EUR 16.6 GBP 21.4 USD 50.3 Pre-diversified Total 100.0 Post-diversified Total 100.0 Diversifciation Benefit as % of Pre-diversified Total Table Y.2: Diversification Results on Property (D&F) Best Estimate Reserve AUD 3.2 CAD 10.0 EUR 8.0 GBP 24.6 USD 54.2 Pre-diversified Total 100.0 Post-diversified Total 100.0 Diversifciation Benefit as % of Pre-diversified Total 99.5th Reserve 11.2 5.3 22.5 28.9 66.1 134.0 122.7 99.5th Reserve 5.5 14.0 14.4 34.5 75.2 143.7 129.9 Capital 3.3 1.6 5.9 7.4 15.8 34.0 22.7 33% Capital 2.4 4.0 6.4 9.9 21.0 43.7 29.9 32% From the tables above, we can see that the diversification benefit is consistent between the two classes modelled. The results are 33% for Casualty and 32% for Property (D&F). Both are also consistent with the theoretical derivation and are higher than the QIS4 cap of 25%. The analysis is on a gross of reinsurance underwriting year basis and has not had catastrophic losses removed. It should not therefore be compared to the proposed standard formula factors which are net of reinsurance and exclude the impact of catastrophes. 18 Appendix 6 Geographical diversification: proposed regional split N. Europe E. Europe N. America C & W Asia W. Europe S. Europe US (see separate map) E. Asia N. Africa Caribbean & C. America S & SE. Asia S. Africa Eastern S. America Oceania NSW S. America 19 Geographical diversification: regional split (US) Mid- west US North- east US Western US South - east US 20