Solvency Assessment and Management: Pillar 1Sub Committee Capital Requirements Task Group Discussion Document 59 (v 3) Life SCR - Lapse Risk

Transcription

1 Solvency Assessment and Management: Pillar 1Sub Committee Capital Requirements Task Group Discussion Document 59 (v 3) Life SCR - Lapse Risk EXECUTIVE SUMMARY This document discusses the structure and calibration of the Lapse risk sub-module of the Life underwriting risk module. The paper includes discussion of the Solvency II developments, consideration of the approaches within other jurisdictions, highlights issues, considers alternatives and recommends an approach going forward for SAM, which incorporates feedback and analysis of SA QIS 2. The task group recommends that an approach based on the Solvency II stresses and general structure be adopted for the Lapse risk sub-module, for comparability across companies and consistency with the way other sub-modules are treated. The general structure of the Lapse SCR sub-module (i.e. three scenarios that apply to lapse up, lapse down and mass lapse events) is considered suitable to capture the key risks relating to lapse-type events within life insurers business. However, based on industry feedback, changes relating to the following key aspects of the Lapse SCR calculation were proposed for SA QIS2 and going forward for SAM: The selective per policy application of the mass lapse and lapse level sensitivities The magnitude and application of the stress applied under the mass lapse sensitivity The aggregation of the mass lapse and lapse level scenarios The rationale behind the methodology followed in SA QIS 2 and proposed for future use in SAM is explained and expanded on in this document. It should be emphasised that the term lapse in the context of this sub-module refers to a wide range of assumptions on the take-up of policyholder options e.g. paid-ups and other options that partially or fully restrict or terminate insurance cover as well as, where relevant, the rate of non-payment of premiums and not taking up options that partly or fully establish, renew or increase insurance cover such as voluntary premium increases. The task group further concludes that the inclusion of a TCAR-style floor to the SCR is not in line with the principles of an economic balance sheet approach on a going-concern basis being developed for SAM, but illiquidity risk should still be assessed. In line with feedback from SA QIS 2 as well as recent developments in Solvency II, it is proposed that the SA QIS 2 approach be retained with the following amendments: The mass lapse parameter for individual business be changed from 45% to 40% Where there is a direct causal relationship between a specific market risk, e.g. equity risk, and policyholder behaviour, allowance for policyholder behaviour should be made within that risk module. More general interrelationships such as the state of the economy and lapse rates should be reflected in the correlation matrices to be used. This is discussed more fully in Discussion Document 48. Companies are not all alike. They have specific characteristics and management practices and will therefore be affected and respond differently to specific stress circumstances. When it comes to lapses, it was considered that the circumstances could vary and could be either company-specific or industry-wide. A split is needed in order to enable companies to consider the types of management actions available to

2 them during extreme scenarios. Some companies may be well suited to dealing with company specific events, but not events that affect the entire industry. It may be the opposite for other companies. Instead of specifying a fixed proportion, a range of ratios is specified which companies should consider when assessing the types of management actions available to them and select the mix that is least favourable to them, with appropriate disclosure. As with other management actions, it can only be allowed if documented and approved by the Board of the company. It is recommended that, for reinsurers, no allowance for diversification between ceding insurers for the portion relating to internal factors be applied to each ceding insurer separately in the mass lapse event. Clarification is proposed to state that the application of the mass lapse event is an instantaneous event at the start of the period (i.e. at valuation date). Homogeneous groups to be used are defined and some clarifications provided. The limitations applied to changes in option take-up (in Solvency II) is considered to determine whether it is still relevant within the South African environment. In particular, it is proposed to remove the maximum downward stress of 20% and to simply subject the stressed option take-up rate to a minimum of 0 in the downward level stress. The task group notes that further work may still be required in future around the following areas: Refinement of the calibration of the stress parameters to reflect South African conditions. A quantitative assessment of this will only be possible if suitable data can be obtained, which is acknowledged to be likely very difficult to achieve given the relatively small and young market. In the absence of evidence or compelling arguments to the contrary, the task group recommends that the mass lapse for retail business risk parameter calibration in Europe for use in Solvency 2 (40%) be followed, and notes that the newest mass lapse stress parameter of 40% for retail business in Solvency 2 does not diverge much from the 45% used in SA QIS 2. Potentially having different mass lapse parameters for different parts of the industry similar to the non-retail and linked business which used a parameter of 70% in SA QIS 2 (allowing for perceived characteristics making it more susceptible to such risks, i.e. institutional investors tend to be more knowledgeable and tend to react more quickly when concerned; and the relatively low costs (surrender penalties) of transferring portfolios between providers). It is recommended that no further split be made, but that developments in this regard in Solvency 2are monitored. Potentially including an allowance for the level of expenses which might be difficult to reduce during the first year or two after a mass lapse scenario. There are arguments for and against this which is considered later in this document, and a wider debate is needed on the topic. There is also a view expressed that the Solvency II calibration of the mass lapse parameters might already include an allowance for this. Currently for SAM it is proposed that no additional requirement on the level of expenses post-mass lapse be included. 1. INTRODUCTION AND PURPOSE This document sets out the recommendations of the Capital Requirements task group with respect to the standard formula capital requirement in respect of lapse risk, where lapse is used in a wider sense to include surrenders, paid-ups and other options altering the cash flows under a policy. This includes non-payment of premiums and not taking up options that have a positive effect on the valuation of a policy, e.g. voluntary premium increase options. Page 2 of 28

3 2. INTERNATIONAL STANDARDS: IAIS ICPs IAIS is the international standards setting body for insurance supervisors. The FSB as a member of the IAIS aims to adhere to these standards. The standards are principled based and set out high level guidance on the setting of solvency capital requirements. There is no reference to detailed capital requirements of individual risk sub-modules such as lapse risk. However, the following are relevant within the broad framework of the capital requirements, of which underwriting risk (and lapse risk as a sub-module) form part (reference: Insurance Core Principles, Standards, Guidance and Assessment Methodology 1 October 2011 ): ICP 17 Capital Adequacy The supervisor establishes capital adequacy requirements for solvency purposes so that insurers can absorb significant unforeseen losses and to provide for degrees of supervisory intervention. Some sub-points in this standard that should be considered includes: 17.7 The supervisor address all relevant and material categories of risk and are explicit as to where risks are addressed, whether solely in technical provisions, solely in regulatory capital requirements or if addressed in both, as to the extent to which the risks are addressed in each. The requirements are also explicit as to how risks and their aggregation are reflected in regulatory capital requirements. Types of risks to be addressed The supervisor should address all relevant and material categories of risk - including as a minimum underwriting risk, credit risk, market risk, operational risk and liquidity risk The supervisor sets out appropriate target criteria for the calculation of regulatory capital requirements, which underlie the calibration of a standardised approach The level at which regulatory capital requirements are set will reflect the risk tolerance of the supervisor. Reflecting the IAIS s principles-based approach, this ICP does not prescribe any specific methods for determining regulatory capital requirements Calibration and measurement error The risk of measurement error inherent in any approach used to determine capital requirements should be considered. This is especially important where there is a lack of sufficient statistical data or market information to assess the tail of the underlying risk distribution. To mitigate model error, quantitative risk calculations should be blended with qualitative assessments, and, where practicable, multiple risk measurement tools should be used. To help assess the economic appropriateness of risk-based capital requirements, information should be sought on the nature, degree, and sources of the uncertainty surrounding the determination of capital requirements in relation to the established target criteria The degree of measurement error inherent, in particular, in a standardised approach depends on the degree of sophistication and granularity of the Page 3 of 28

4 methodology used. A more sophisticated standardised approach has the potential to be aligned more closely to the true distribution of risks across insurers. However, increasing the sophistication of the standardised approach is likely to imply higher compliance costs for insurers and more intensive use of supervisory resources (for example, in validating the calculations). The calibration of the standardised approach therefore needs to balance the trade-off between risk sensitivity and implementation costs. 3. EU DIRECTIVE ON SOLVENCY II: PRINCIPLES(LEVEL 1) Relevant extracts from the Solvency II level 1 principles are provided below. As is the case with the IAIS core principles, these requirements are in nature of a higher level than required for the establishment of detailed requirements of the lapse risk sub-module of the Life Underwriting risk module within the capital requirements. However, it provides the broad framework within which these requirements are to be considered. Article 101 Calculation of the Solvency Capital Requirement 3. The Solvency Capital Requirement shall be calibrated so as to ensure that all quantifiable risks to which an insurance or reinsurance undertaking is exposed are taken into account. It shall cover existing business, as well as the new business expected to be written over the following 12 months. With respect to existing business, it shall cover only unexpected losses. It shall correspond to the Value-at-Risk of the basic own funds of an insurance or reinsurance undertaking subject to a confidence level of 99,5 % over a one-year period. 4. The Solvency Capital Requirement shall cover at least the following risks: (a) non-life underwriting risk; (b) life underwriting risk; (c) health underwriting risk; (d) market risk; (e) credit risk; (f) operational risk. 5. When calculating the Solvency Capital Requirement, insurance and reinsurance undertakings shall take account of the effect of risk-mitigation techniques, provided that credit risk and other risks arising from the use of such techniques are properly reflected in the Solvency Capital Requirement. Article 104 Design of the Basic Solvency Capital Requirement 1. The Basic Solvency Capital Requirement shall comprise individual risk modules, which are aggregated in accordance with point (1) of Annex IV. It shall consist of at least the following risk modules: (a) non-life underwriting risk; Page 4 of 28

5 (b) life underwriting risk; (c) health underwriting risk; (d) market risk; (e) counterparty default risk. 2. For the purposes of points (a), (b) and (c) of paragraph 1, insurance or reinsurance operations shall be allocated to the underwriting risk module that best reflects the technical nature of the underlying risks. 3. The correlation coefficients for the aggregation of the risk modules referred to in paragraph 1, as well as the calibration of the capital requirements for each risk module, shall result in an overall Solvency Capital Requirement which complies with the principles set out in Article Each of the risk modules referred to in paragraph 1 shall be calibrated using a Value-at-Risk measure, with a 99,5 % confidence level, over a one-year period. Where appropriate, diversification effects shall be taken into account in the design of each risk module. 5. The same design and specifications for the risk modules shall be used for all insurance and reinsurance undertakings, both with respect to the Basic Solvency Capital Requirement and to any simplified calculations as laid down in Article With regard to risks arising from catastrophes, geographical specifications may, where appropriate, be used for the calculation of the life, non-life and health underwriting risk modules. Article 105 Calculation of the Basic Solvency Capital Requirement 3. The life underwriting risk module shall reflect the risk arising from life insurance obligations, in relation to the perils covered and the processes used in the conduct of business. It shall be calculated, in accordance with point (3) of Annex IV, as a combination of the capital requirements for at least the following sub-modules: (f) the risk of loss, or of adverse change in the value of insurance liabilities, resulting from changes in the level or volatility of the rates of policy lapses, terminations, renewals and surrenders (lapse risk); Article 109 Simplifications in the standard formula Insurance and reinsurance undertakings may use a simplified calculation for a specific sub-module or risk module where the nature, scale and complexity of the risks they face justifies it and where it would be disproportionate to require all insurance and reinsurance undertakings to apply the standardised calculation. Simplified calculations shall be calibrated in accordance with Article 101(3). 4. MAPPING ANY PRINCIPLE (LEVEL 1) DIFFERENCES BETWEEN IAIS ICP & EU DIRECTIVE Page 5 of 28

6 No differences the EU Directive is in line with the IAIS core principles. 5. STANDARDS AND GUIDANCE (LEVELS 2 & 3) 5.1 IAIS standards and guidance papers This was covered in section 2 above CEIOPS CPs (consultation papers) The level 2 advice (formerly CP49) draws a distinction between three very different areas where allowance is made for lapse (and other options) risk: 1. Within all the other modules, where a scenario is used to determine the capital requirements, in order to capture the complex dependency between option take-up behaviour and other risks within the standard formulae. Examples include allowance within the market risk capital requirements for the effect of management actions on surrenders/lapses; the effect of the takeup of a lump sum vs. annuity in differing interest rate scenarios; etc. 2. Lapses triggered by worry because of deterioration in the financial position of the re-/insurer etc. 3. Misestimate of current lapse rates given the current scenario. The lapse risk sub-module is intended to cover the risks in 2 and 3, while the interaction risks within 1 is to be captured within the other separate risk-modules, by using assumptions regarding option take-up behaviour which makes sense in the relevant scenario. In addition, a wide definition is applied to the term lapse, so that not just the options which would fully terminate or renew a policy is included, but also for example paidup risk. The following paragraphs give these definitions: Under the wide definition of scope, the scenarios which define a permanent decrease and increase of option take-up rates could be defined as follows: lapse shock down = Reduction of x% in the assumed option take-up rates in all future years for all policies without a positive surrender strain or otherwise adversely affected by such risk. Affected by the reduction are options to fully or partly terminate, decrease, restrict or suspend the insurance cover. Where an option allows the full or partial establishment, renewal, increase, extension or resumption of insurance cover, the x% reduction should be applied to the rate that the option is not taken up. lapse shock up = Increase of y% in the assumed option take-up rates in all future years for all policies with a positive surrender strain or otherwise adversely affected by such risk. Affected by the increase are options to fully or partly terminate, decrease, restrict or suspend the insurance cover. Where an option allows the full or partial establishment, renewal, increase, extension or 1 The IAIS Insurance Core Principles, Standards, Guidance and Assessment Methodology issued October 2011 has superseded previous Standards and Guidance (in this case Standard No and Guidance paper No on the structure of regulatory capital requirements). Page 6 of 28

7 resumption of insurance cover, the y% increase should be applied to the rate that the option is not taken up The surrender strain of a policy is defined as the difference between the amount currently payable on surrender and the best estimate provision held. The amount payable on surrender should be calculated net of any amounts recoverable from policyholders or agents e.g. net of any surrender charge that may be applied under the terms of the contract. Capital requirements for the three sub-risks (see paragraph 3.142) should be calculated based on a policyby-policy comparison of surrender value and best estimate provision. In this context, the term surrender should refer to all kind of policy terminations irrespective of their name in the terms and conditions of the policy. In particular, the surrender value maybe zero if no compensation is paid on termination. The calculation of the capital requirement is based on three scenarios: Permanent increase in lapse rates Permanent decrease in lapse rates Mass lapse event The Capital requirement for lapse risk is obtained as the loss of net asset value (per policy) under the most adverse of the three scenarios. The advice acknowledges that using the maximum requirement from three different scenarios (permanent increase in rates, permanent decrease in rates, and a mass lapse event) has some shortcomings. This simple approach has some shortcomings. For example, an insurer may be exposed to the risk of an increase in lapse rates on one part of its portfolio and a decrease of lapse rates in another part of the portfolio. Such situations are not covered by the approach. However, within the natural limitations of the standard formula the QIS4 approach appears to be an acceptable solution. The feedback from the QIS4 participants gives no other indication. In particular, one could be exposed to a portfolio that currently have policies behaving differently to each of these scenarios, but is perhaps only exposed to one later on which might cause the need for a sudden increase in capital requirements (take a new-age risk product with a mix of younger (negative reserve) and older policies (positive reserve) for example) Permanent increase/decrease calibration: Calibration for QIS4 was done using a study from the UK with-profit Life insurance market in 2003 performed by the FSA. Quantiles produced in the study were lower than the confidence level required for Solvency II, but by way of extrapolation a stress of -50% was proposed. Using an argument of symmetry, the upward stress was taken as 50%. An analysis of the Polish supervisor shows that the 99.5 th quantile of annual lapse rate deviations from a long term mean is between 60% and 100% for increases and - 60% and -90% for decreases. These are based on annual deviations and as such overestimate the risk of a permanent change in rates. However, they do seem to support the rates of 50% and -50%. Some restrictions were advised such that shocked rates do not exceed 100%, and do not change by more than 20% in absolute terms in the downward direction. Page 7 of 28

8 Mass lapse event calibration: This could have different causes e.g. changes in tax regulation, deterioration of financial position of insurer. This is comparable to a run on the bank scenario and is an immediate, drastic rise in lapse rates. The calibration should not take account of such scenarios triggered by events in other risk modules (where it should be allowed for as described above within that module, e.g. equity risk). It should allow for the fact that this is a catastrophe type event and for herd behaviour. Under Solvency II, technical provisions may be less than the policy s surrender value and this could pose a risk to insurers with large exposure to such policies, especially should this become public. Empirical data for calibration for such an event is very scarce. In the absence of better evidence, CEIOPS proposed to maintain the QIS4calibration of 30% of the sum of positive surrender strains for retail policies, noting that many policies have guarantees and other features e.g. surrender penalties which improve persistency. However, for non-retail business (including pension fund management business) where institutional investors tend to be better informed and where there are generally no surrender penalties, a higher stress factor is merited. CEIOPS proposed the QIS3 calibration of 70% of the sum of positive surrender strains. Simplifications Using the principle of proportionality, it can be considered acceptable to use homogeneous risk groups instead of a policy-by-policy approach, with the following criteria for using it: As the calculation on a homogeneous risk group level is likely to result in the same or a lower capital requirement than the policy-by-policy calculation, it seems necessary to set up criteria for its application. A calculation on the level of homogeneous risk groups should be considered to be proportionate if: (a) the homogeneous risk groups appropriately distinguish between policies of different lapse risk; (b) the result of a policy-by-policy calculation would not differ materially from a calculation on homogeneous risk groups; and (c) a policy-by-policy calculation would be an undue burden compared to a calculation on homogeneous risk groups which meet criteria (a) and (b). A further simplification is provided below which was used in QIS 4 and QIS 5: Factor-based formula for scenario effect For the two scenario calculations lapse shock down and lapse shock up, factor-based simplifications were provided in QIS4. These simplifications attempt to approximate the effect of the permanent change of lapse rates by projection of the effect of a temporary shock into the future. The effect of a temporary change in lapse rates can easily be measured by means of the surrender strain: for example, the loss incurred in a portfolio with positive surrender strain due to temporary change of lapse rates by 5 percentage points is approximately 5% of the surrender strain. In order to for the permanence of Page 8 of 28

9 the change in the scenarios lapse shock down and lapse shock up, this loss can be multiplied with the duration of the portfolio in question This approach results in formulas as follows: Lapse down = 50% l down n down S down and Lapse up = 50% l up n up S up, where l down ;l up = estimate of the average rate of lapsation of the policies with a negative/positive surrender strain n down ;n up = average period (in years), weighted by surrender strains, over which the policy with a negative/positive surrender strain runs off S down ;S up = sum of negative/positive surrender strains The simplified calculation should be done at an appropriate granularity The factor-based approximations should only be used if they are proportionate to the nature, scale and complexity of the risk. In particular, as an application of the scale criterion of the proportionality principle, the simplification should only be used if the capital requirement for lapse risk(determined with the simplification) is small compared to the overall capital requirement. A threshold of 5% of the overall SCR (before adjustment for the loss-absorbing capacity of technical provisions and deferred taxes) was tested in QIS4 and appears to be appropriate. Moreover, the simplification should only be used, if the more sophisticated result of the scenario analysis is not easily obtainable. 5.3 Level 2 Implementing Measures (Draft 31 Oct 2011) The main changes from CP49 (and QIS5) are: An increase in the mass lapse shock calibration for retail business from 30% to 40% Addition of a 3 rd category for mass lapse, namely reinsurance contracts covering (re)insurance contracts that will be written in the future a decrease of 40 % of the number of those future (re)insurance contracts used in the calculation of technical provisions to be applied The latest technical specification 2 released by EIOPA also includes slightly more detail for the simplification (indicated in bold below): Lapse down = 50% l down n down S down and Lapse up = 50% l up n up S up, where l down ;l up = estimate of the average rate of lapsation of the policies with a negative/positive surrender strain, subject to a minimum rate of lapsation of 40% in case of negative surrender strain and a minimum of 67% in case of positive surrender strain 2 EIOPA-DOC-12/362: Technical Specifications for the Solvency II valuation and Solvency Capital Requirements calculations (Part I), 18 October 2012 Page 9 of 28

10 n down ;n up = average period (in years), weighted by surrender strains, over which the policy with a negative/positive surrender strain runs off S down ;S up = sum of negative/positive surrender strains 5.4 Other relevant jurisdictions (e.g. OSFI, APRA) The Canadian regulator (OSFI) has not yet published anything on the envisaged requirements in terms of underwriting risk. Only information on market risk is available. The Australian approach to capital requirements for lapse risk is as follows: LPS 3.04 (Capital Adequacy Standards) requires life insurers to recalculate their liabilities using stressed assumptions, and to hold the difference between best estimate liabilities and stressed liabilities as a capital requirement. The stresses are expressed in the form of minima, as well as high estimates there is nothing to prevent an insurer from using something above the high figure, but it gives insurers a sense of what APRA believes is a reasonable range of stresses. For lapse risk, the minimum margin to be applied is 25%, while the high estimate is 100%. APRA also takes into consideration a TCAR-style requirement, where the minimum capital to be held in respect of a policy or group of policies is equal to the Current Termination Value of those policies. The CTV makes allowance for policyholders reasonable expectations. 5.5 Mapping of differences between above approaches (Level 2 and 3) The approach in Solvency II seems to have the most comprehensive treatment of lapse risk. The TCAR requirement probably attains something similar to the treatment of lapse-related issues within the Own Funds part of the Solvency II regulation. This treatment is not in the scope of this paper. 6. ASSESSMENT OF AVAILABLE APPROACHES GIVEN THE SOUTH AFRICAN CONTEXT 6.1 Discussion of inherent advantages and disadvantages of each approach The Solvency II approach is relatively simple to apply (3 scenarios). In addition, appropriate simplifications are provided. The approach to include the option take-up behaviour within other risk modules is a major advantage to ensure that nothing is missed relating to interdependencies with other risks, especially since the current structure and correlations doesn t lend itself to modelling these complex interactions and strong correlations well. There is some debate as to the appropriateness of selective lapse stresses as specified in the current Solvency II methodology. This is discussed in section below. The possibility of a mass lapse event caused for instance by reputational risk should not be excluded and it is therefore proposed that the mass lapse shock be retained. Page 10 of 28

11 6.2 Impact of the approaches on EU 3rd country equivalence There should not be an impact at this level of granularity as long as the main principles are adhered to. 6.3 Comparison of the approaches with the prevailing legislative framework Currently, for life insurers there are compulsory margins (10% for surrenders and 25% for lapses) up or down from the best estimate lapse rates on a product level (whichever results in a higher reserve for each future time period). In addition, there is a minimum CAR requirement which ensures that the total of statutory reserve PLUS CAR for each policy cannot be less than its surrender value (which may be 0) (TCAR). The aim of TCAR was to ensure that insurers could survive a worst case run on the bank scenario, and not treat any policy as an asset. Alternatively, this can be viewed as the amount that will be required on a wind-up as opposed to a going-concern basis. Furthermore, a portion of this TCAR is included in the more risk-based calculation of the OCAR. This is calculated as 40% of the amount needed to ensure no policy has a negative liability (in the case of lapses) and 20% of the amount needed to ensure no policy s liability is less than its surrender value (in the case of surrenders). The lapse and surrender component of OCAR was chosen to provide for roughly a doubling of lapse or surrender rates. (Note that CAR is targeted at a 95% confidence level over 5 years.) It has been debated within the SAM task groups whether something similar to TCAR should be included in the Lapse SCR, given negative reserves being allowed within technical provisions under SAM. It should be noted that the issue of TCAR within the current CAR formula has previously been debated by the CAR sub-committee of the Actuarial Society in a 2003 paper 3. The following issues were noted in respect of the TCAR approach: There is widespread acknowledgement that the TCAR is contrary to the notion of a capital requirement that gets insurers to the 95th percentile, in that it is an extreme run-on-the-bank scenario. The notion of taking the FSV reserve and the CAR together to essentially leave a company with no negative reserve also appears to be contrary to the basic notion of the IAA Working Party 4 that the capital requirements are not established by looking back at the method and basis used for basic reserving purposes. There is an argument that really the TCAR minimum is imposed to ensure that in effect no insurer carries a negative reserve as an asset. However, there are a number of issues around this principle. First, if it is agreed that the IAA Working Party s view that solvency regulations should be established so that capital requirements genuinely capture true underlying risk and should not be dependent upon the accounting method and basis used, then the current approach of essentially zeroing out negative reserves by combining the FSV reserve and the CAR is contrary to such. If the intention is to ensure that no insurer carries a negative reserve, then this is a reserving issue. 3 Capital Adequacy for Long-Term Insurers Revisited (Dardis et al 2003) 4 A Global Framework for Insurer Solvency Assessment. IAA Insurer Solvency Assessment Working Party, Draft 24, May 2003 Page 11 of 28

12 In spite of the above arguments against TCAR, the sub-committee recommended that TCAR be retained, mainly because any change in this regard would have required changes to existing legislation. From a SAM point of view, the Capital Requirements task group noted that retaining TCAR would be counter to the economic balance sheet approach on a going-concern basis being developed for SAM, and that under this approach there was no economic reason to hold TCAR. Furthermore it was considered unlikely that a 100% selective mass lapse event would constitute a 1-in-200 year event. Appropriate calibration of the Solvency II approach should make proper allowance for the mass lapse risk and the probability of losing negative reserves. However, illiquidity risk is likely to be dealt with by way of additional disclosure which is outside the scope of this document. 6.4 QIS5 report Lapse risk was the largest component of the Life risk module, comprising 46% of the undiversified Life SCR of solo life entities. EIOPA s QIS5 report 5 noted that Life underwriting risk was generally well received with the exception of Lapse risk, where the following main concerns were raised (extracted from the QIS5 report, with our emphasis added): The key practical criticism was the need to calculate lapses on a policy-by-policy basis: a large number of undertakings raised this as an area in need of simplification, and generally they were supported by their supervisors. Criticisms were that this was too onerous in terms of calculation time (especially for complex or stochastic models) and that new systems will have to be developed at significant cost. In some cases ad hoc simplifications were performed, or model points were used. There were also criticisms from undertakings and some supervisors of the policy-bypolicy approach on more theoretical grounds, with some suggesting that the treatment of surrender strain should not be asymmetric and should be by broad segment to better reflect lack of policyholder rationality. Some said that taking the maximum of the three shocks was insufficient, and that a more subtle approach should be applied, and some questioned the dividing line between wholesale and retail business, usually because it caught the wrong business (rather than due to the concept of the division itself). 6.5 SA QIS1 report As in Europe, Lapse risk in QIS1 was the largest component of Life underwriting risk, comprising just over 60% of undiversified Life SCR. The mass lapse scenario bit in 50% of the submissions. The QIS1 report 6 noted the following industry feedback in respect of Lapse risk: There was uncertainty regarding the treatment of expenses under the lapse scenario, especially the mass lapse scenario. There was also a concern that it was too onerous to calculate the lapse exposure on a policy by policy basis, as already required in the current CAR capital calculations. Some insurers had concerns that the structure of the lapse risk calculation was too conservative, as the combination of upwards and downwards shocks [imply] that policyholders will always be able to select against the insurer. 5 EIOPA Report on the fifth Quantitative Impact Study (QIS5) for Solvency II 6 Report on the results of 1st South African Quantitative Impact Study ( SA QIS1 ) Page 12 of 28

13 However, other insurers highlighted that both mass lapse as well as a movement up or down should be included, to ensure that all lapse risks are covered in the capital requirement With regard to the parameters used, there were concerns that the considerable difference in the mass lapse stress between retail business (30% mass lapse) and nonretail business (70% mass lapse) was too large and not justifiable. There was also a concern that the lapse shocks applied to life underwriting were different to the lapse shocks applied to the health underwriting module, especially for products that have both life and health benefits. 6.6 Adjustments to the lapse SCR sub-module for SA QIS2 Following the feedback from SA QIS1 and other industry comments, it was decided that a number of adjustments are required to the definition of the lapse SCR submodule. The adjustments relate to the following: The selective per policy application of the mass lapse and lapse level sensitivities The magnitude and application of the stress applied under the mass lapse sensitivity The aggregation of the mass lapse and lapse level scenarios The following paragraphs discuss each of the above adjustments in detail Selective per policy application of the mass lapse and lapse level sensitivities The selective per policy application of the lapse level and mass lapse sensitivities relies on two assumptions: Policyholders have access to information relating to the economic value of their policy to the insurer Policyholder behaviour is driven by the economic value of their policy to the insurer, rather than by the economic value of the policy to themselves (The two can be very different). In practice, policyholders do not have access to the type of information in the first bullet and it is unlikely that they will gain access to such information under a stress scenario. Also, while conceptually it is possible that policyholders behaviour will be influenced by the economic value of their policy, it is more likely that their behaviour is more sensitive to extreme economic and reputational events. The insurer could also have mis-estimated lapse rates, but this in any case implies a mis-estimation on a group of policies rather than single policies. On the other hand, it could be possible that lapse experience will differ between product lines under extreme circumstances. For example: policies with contractual guarantees may be less likely to lapse under economic extremes (or reputational events) compared to policies without contractual guarantees Level premium risk products may be less likely to lapse compared to risk products where premiums are age rated, etc. Given the above considerations, it was decided that while the selective per policy application of lapse sensitivities had significant conceptual flaws, there could be a case for lapse sensitivities to be applied selectively to homogenous risk groups and Page 13 of 28

14 the definition for SAQIS2 was amended accordingly. The possibility of a mass lapse event caused for instance by reputational risk should not be excluded and it is therefore proposed that the mass lapse shock be retained. For SA QIS2 a homogenous risk group was defined as the level at which lapse assumptions are set in the calculation of the best estimate component of technical provisions Magnitude and application of the mass lapse event stress The SA QIS1 mass lapse shock of 30% for retail business and 70% for non-retail (corporate) business was based on the Solvency II QIS5 calibration. Following feedback from the industry on the calibration of the mass lapse submodule it was decided that the 30% shock in respect of retail business was too low to represent a 1-in-200 year event. This was in light of relatively high base lapse rates experienced by insurers on some product lines (e.g. up to 20% on some investment contracts in their first year since inception). A call was made for a more reasonable ( higher ) calibration of this sub-module. Due to lack of data relating to mass lapse events it was not possible to perform a calibration of a 1-in-200 year event. To test the impact of a more reasonable mass lapse event, it was therefore decided to apply a stress of 45% for SA QIS2, which is 50% higher than that applied in SA QIS1. Interestingly enough, the latest Solvency II specifications have also increased the size of the mass lapse stress to 40%. For SA QIS2 it was also decided to include a more locally relevant split for the retail and corporate mass lapse stress, which should be based on the characteristics of the underlying business rather than the type of insurer. The following key differentiation was considered appropriate: Business where the policyholder can lapse their policies or switch their policies to other providers without a significant loss in value should apply a mass stress of 70%. This would include corporate business and linked business. Business where the policyholder cannot lapse their policies or switch their policies to other providers without a significant loss in value should apply a mass stress of 45%. For SA QIS2 it was therefore proposed to replace the SA QIS1 (i.e. QIS5) definition of retail/non-retail by the following: Group and Grouped Individual business - 70% stress Linked business (group and individual) 70% stress Individual business (excluding individual linked business) 45% stress In order to provide objective definitions of the above groupings it was decided to use the definitions included in the current Long-term Insurance Act (LTIA) and Guidance to the LT returns, as follows: Group Business and Grouped Individual Business is defined as follows 7 : 7 Guidance Manual for the completion of the Long-term Insurance Statutory Returns Page 14 of 28

15 Group Business: Insurance where a long-term policy is issued to a policyholder other than an individual that covers a group of persons identified by reference to their relationship to the entity buying the contract provided this excludes grouped individual business. Grouped Individual Business: Insurance where a long-term policy is issued to a policyholder other than an individual. In terms of the policy an identifiable individual(s) or member(s) is the life insured(s). Only the individual(s) or member(s) may terminate the cover. Linked business is defined as follows 8 : Business that relates to liabilities under a linked policy, where a linked policy means a long-term policy of which the amount of the policy benefits is not guaranteed by the long-term insurer and is to be determined solely by reference to the value of particular assets or categories of assets which are specified in the policy and are actually held by or on behalf of the insurer specifically for the purposes of the policy. Individual business is defined as follows: Insurance where a policy is issued to an individual (excluding linked individual policies, which are included under Linked business above) The aggregation of the mass lapse and lapse level scenarios In SA QIS1 the overall lapse SCR was defined as the maximum of a level increase/decrease in lapses event and a mass lapse event, implying that the two events are mutually exclusive. However, in practice it is possible for the two events to take place simultaneously. (For example, a company may suffer a reputational event upon which it experiences mass lapses on a significant portion of its in-force book. This could then lead to operational difficulties for the company resulting in a further increase in long term lapse expectations. On the other hand, it could also be argued that if people have not lapsed their policies under the mass event they are also less likely to do so in the future.) It was therefore felt that the SA QIS1 definition understates the size of the overall lapse risk and it was decided that the lapse level and mass lapse events should be aggregated. In aggregating the mass lapse and lapse level events, the task group noted that the occurrence of the one event will offset the financial impact of the other. For example if a mass lapse event occurs resulting in a significant reduction in the in-force book of business, then the financial impact of the level lapse event will only relate to the business that remains in-force. If the mass lapse and level lapse events are 100% correlated, then the above can be expressed using the following equation: 8 Long-term Insurance Act 52 of 1998, as amended by the Insurance Laws Amendment Act 27 of 2008 and the FSB directive 146.A.i (LT) Page 15 of 28

16 ( ) (( ) ) Where is the current mass lapse stress NAV impact is the current lapse level stress NAV impact (higher of lapse up or lapse down stress) is the size of the mass lapse stress However if the correlation is less than 100%, say if the events were independent (0% correlation), then applying a factor of (1 - ) will exaggerate the reduction in the level lapse event and result in an understated Lapse SCR. Under the independence assumption, the occurrence of a 1-in-200 year level lapse event does not imply a simultaneous 1-in-200 year mass lapse event. The mass lapse event may be much smaller (say 1-in-10 years) or not happen at all. This means that the reduction in the financial impact of the level lapse event may be a lot less than that applied through the (1- ) factor (i.e. if no mass lapse event occurs there should be no reduction to the impact of the level lapse event). An investigation was performed to find a solution to the above problem. The two approaches considered were: Calculate and introduce a small negative correlation (say ρ ) between the mass and level lapse events Reduce the impact of the (1- ) factor, by introducing a factor (say x ), such that (1- )< (1 x* ) The investigation assumed independence between the level lapse and mass lapse events. It also defined distributions for the mass and level lapse events calibrated such that the 99.5th percentile corresponded to a 45% mass and 50% level lapse shocks (as planned for QIS2). From these distributions were generated random events for the level, mass and combined mass and level events and the worst 99.5th percentile event was extracted for each. Values for ρ and x were calculated to find a combination of the mass and level lapse events that will be as close as possible to the simulated combined mass and level lapse event. The calculations were performed numerous times for different standalone impacts of mass lapse and level lapse on a hypothetical insurance company. From the analysis it was possible to draw the following conclusions: The value of ρ that results in the most reasonable aggregated value of the mass and level lapse events is roughly -22%. This was the best fit overall for the various impacts of mass and level lapse events but could result in some counterintuitive results (i.e. a combined mass and level lapse event that is less than the mass lapse or level lapse events on their own) Page 16 of 28

17 To mitigate the above a maximum can be applied to the level, mass and aggregated events. Introducing a negative correlation into the aggregation formula may be difficult to interpret and inconsistent with the interpretation of the correlations between the various other risk types. The value of x that results in the most reasonable aggregated value of the mass and level lapse events is roughly 0.5. This could also result in counterintuitive results in cases where the impact of the mass lapse event is significantly less than the level lapse event, in which case the aggregated result will be less than the level lapse event on its own. To mitigate the above a maximum can again be applied to the level, mass and aggregated events. After considering the above it was decided that the x factor of 0.5 should be applied to the lapse SCR formula for SA QIS2. It was also decided to maintain the maximum of the level, mass and aggregated lapse impacts. Given the lack of data it was also decided that the independence assumption was not unreasonable. The formula for the basic aggregated lapse event was therefore defined as: (( ) ) And, the overall lapse SCR formula can then be defined as: ( ) Some additional adjustments were then made to the above formulae as follows: Similar to the level lapse events, the mass lapse event was also to be applied selectively per homogenous risk group (subject to a floor of 0).For a homogenous risk group where the mass lapse impact is nil, is set to 0. Although and are calculated per homogenous risk group, the above aggregations and maxima apply to the business as a whole (and not per homogenous risk group). The alternative is to apply the aggregation and maxima to each homogenous risk group, resulting in a Lapse mass event applying to one group, a level lapse event to another group, an aggregated event to a third group and so on. It was decided that such an event was not plausible and the approach was therefore disregarded. All of the above was taken into account into the final SA QIS2 lapse SCR definition, and the formula was expressed as follows: Page 17 of 28

18 6.7 Feedback and debate following SA QIS 2 The following are highlighted in the SA QIS 2 report and relevant to this discussion document: Life underwriting risk is the largest component of the BSCR. In fact, Lapse risk was the largest risk included in the SCR for SA QIS 2, increasing from 69% of total life underwriting risk to 79%. The following is taken from the SA QIS2 report: Due to all the changes made to the lapse sub-module, lapse risk is now the sub-module in the standard formula which generates the highest capital requirement. This large capital requirement reflects the extent to which positive net cash-flows are taken into account in the valuation of technical provisions. The table below sets out the results of applying the different stresses described above. Table 6.6: Results of the various lapse stresses required under SA QIS2 From the table above we can see that the level shock, given a mass shock is the most onerous of the stresses tested, with 34 of 38 insurers finding this the most onerous stress. The three insurers where the level shock was more onerous are all reinsurers where a decrease in the level of lapses led to the most onerous capital requirement. There were a number of concerns raised at the increased level of the mass lapse shock, as well as the requirement to calculate the level shock, given a mass shock. There was some concern that the latter stress resulted in double-counting of the lapse risk. In addition to the calculations above, insurers were asked to perform a sensitivity calculation to consider the impact of keeping total expenses unchanged for a period of two years after the mass lapse event, which would lead to higher per policy expenses over the period. The application of this stress leads to a 20% increase in the capital requirement from the mass lapse stress where insurers do not change their per policy expenses. Page 18 of 28