Board for Actuarial Standards

MEMORANDUM To: From: Board for Actuarial Standards <basinsurance@frc.org.uk> Chaucer Actuarial Date: 20 November 2009 Subject: Chaucer Response to BAS Consultation Paper: Insurance TAS Introduction This memo forms Chaucer Actuarial s response to the Insurance TAS consultation paper issued by the Board for Actuarial Standards in September 2009. Chaucer Response a) One TAS or Two? It is our opinion that there should be two TASs - one for general insurance and one for long term insurance. Whilst many principles are shared between general insurance and long term insurance, these will be covered in the generic TASs in Reporting, Modelling and Data which apply across the range of actuarial work. When considering the specifics of general insurance and long term insurance, i.e. those items which would not be covered in the generic TASs, there can be very different considerations. By covering both general insurance and long term insurance in the same TAS, there is a danger the TAS includes too much information not relevant to the other area. Alternatively, there is a danger that the principles are so generic that little is added beyond what is in the generic TASs. As an example, for long term insurance, key assumptions would include mortality and morbidity. These concepts have no relevance to general insurance. At the same time it would be very odd if the specific TAS did not include comments on these. Having two TASs has benefits when future changes are required which affect only one of the areas. There will be a time saving as it will be very clear to individuals whether any change affects them. This may not be so clear if both long term business and general insurance are covered under the same TAS. An additional environmental benefit will be that less paper is used as only those working in the area where the TAS has changed will need to print out a new copy! b) Purpose The purpose as set out seems reasonable. If this is applied in the correct areas (if the scope of this TAS is fit for purpose) and the principle of proportionality is applied then the purpose will be (and should be) fulfilled. We believe that the materiality of decisions being taken by an organisation needs to be taken into account. The purpose is reasonable when applied to significant decisions in the context of a whole entity but not to the numerous minor decisions made which are immaterial in the context of the whole entity. Another issue here is that the requirements should not be so onerous that the fulfilling of the requirements delays important decisions or prevents the actuary from working on other equally

important work. If the scope is too broad then the actuary will be prevented from supporting the entity in its key decisions and could lead to more work being done by less qualified, non actuaries, neither of which should be a desirable consequence of this TAS. c) General Concepts As mentioned above the principle of materiality should not only be applied to the matters pertaining to a decision but to the importance to the organisation of the decision itself. Contrast the importance of a full restructure of a reinsurance program for an entity as opposed to the isolated pricing of a single reinsurance layer. Only the first one should be considered material and the second should be out of scope. Where we exclude work from scope under the principle of proportionality are we expected to justify each one? The TAS should address this issue. d) Scope - Introduction and Rationale Much of our work is done as a combination of efforts of actuaries and non actuaries. Where the contribution to reports from non-actuaries is significant, the application of this TAS is likely to be impractical. One of the fundamental issues with TAS R and consequently anything that is brought into scope in TAS I appears to be the requirement that for a report to comply it needs to be provided before the decision is made. This, in practice, pushes the actuary to provide a report before a decision is made through their desire to satisfy the professional guidance. This has the following issues: o o Often this is entirely impractical given the timescales involved and the amount that s required to satisfy the generic TASs (including the Data and modelling TASs).; Where a piece of actuarial work is an opinion on a decision then this necessitates the report to be written prior to the decision being made which may be impractical and restrict the ability for the actuary to make any such comment post the decision. Actuaries left with the judgement about what is and what is not useful for the company in the context of these requirements. In many cases the company may not want report after report on reserving, pricing, business planning or capital modelling decisions and may make this very clear to the actuary. In these instances the actuary should be permitted to take this into consideration when deciding on how to apply these standards. e) Scope Reserving and Reporting (incl. related other comments) 2.8 refers to cashflows as particularly important which is not true in most actuarial GI work. We believe this is an example of how a combined TAS for GI and Life doesn t work. Section 4.28 gives some cause for concern, where scope is extended to planning. It will be difficult to strip out certain components of this process for TAS application. Sections 4.33 4.40 open up a whole can of worms. Reports on due diligence / new class opportunities pretty much through necessity, given the real paucity of data available probably fall short of what is required going forward, so exclude from scope. 5.13 Factors extraneous to the insurer e.g. economy, legislation etc. Clearly GI is greatly affected by several such factors, but how is a whole different issue. Whilst it s pretty easy to pick up and document external data on such factors, how they affect our liabilities etc. is not a simple extrapolation. This, therefore should not be part of the requirements. 5.20 suggests that where data is poor, the actuary should modify the data. In GI, data is nearly always poor, and it is difficult to modify the data as you usually get it at a consolidated level. The usual approach would be to intelligently use this data which could mean using higher or lower assumptions than the pure data implies. This is not the same as incorporating margins within assumptions, we would always strive for a best-estimate, even where data is poor/limited.

Section 6.20 may need clarifying. It refers to material events after the effective date of data but this will include events after the valuation date itself. Is this the intention? There is a section (7.32) that suggests looking at past and possible future variations in currency rate of exchange. We don t think this would be productive for us the variations in past and possible future exchange rates are (1) going to be difficult, if not impossible to split out anyway, and (2) are often not material when viewed in the context of the overall volatility of GI business. An actuary is not necessarily an expert in exchange rate volatility and should be permitted to exclude from scope if appropriate. f) Scope - Pricing We do not believe pricing should be included in the scope for TAS I. The product range in general insurance is vast and many companies will write a great number of different product lines. This makes compliance with the TAS extremely onerous, even impossible. The principle of proportionality applied to each individual piece of pricing work would appear to naturally exclude each such pieces of work, making bringing this work in scope irrelevant. Compliance with the TAS in such a situation could significantly increase costs. This may result in actuarial input being sought less for pricing work. There is no requirement for pricing work to be carried out by an actuary and pricing work may be carried out by other technically trained individuals. The increased cost of complying with the TAS may well lead some companies to prefer to employ non-actuaries to do their pricing. g) Scope Capital Modelling The relevant areas to capital modelling actuaries that are proposed to be included in the scope of TAS I are: 4.73 e) assessing regulatory capital requirements; 4.73 g) actuarial information used in product design and pricing; 4.73 h) actuarial information used in business planning; 4.73 k) work related to opining on underwriting policy and reinsurance arrangements We agree that e) should be included, although it should be noted that regulatory capital is not simply calculated by actuaries but by a multi-disciplinary team. Indeed the parts done by actuaries draws on work done by many non actuaries within the business. Bearing this in mind, how should this work in practice? Should the scope be restricted to just that part that of the regulatory capital calculation that he actuaries do and how should the dependency on non actuaries for information and expert judgement be addressed? K) should not be included. The decisions will often pre-date the actuarial function opinion work. How can it be TAS R compliant in this case? We agree with the sentiment of including g) within TAS I, although we think that establishing whether an actuary had followed the standards in product design and pricing work (i.e. separating the information and the decision) would be difficult in practice, so exclude from scope. See pricing section later. For h) we think that distinction should be made as to which specific actuarial inputs to business planning are covered under TAS I. Actuarial input to business planning can include claims reserving, premium rating assessments, capital allocation and testing capital requirements of different portfolios. While some of this probably should be subject to the TAS I standards (e.g. claims reserving), we think that the documentation, data and assumption requirements are too onerous for many of the inputs that a capital modelling actuary would typically have to business planning (see comments about 4.76a below). We suggest that making TAS I apply to actuarial information of past business performance used in business planning would make clearer the distinction between 4.73 h) and 4.76 a) The relevant areas to capital modelling actuaries that are proposed to be not included in the scope of TAS I are:

4.75 a) decisions in business planning, product design and pricing 4.75 b) investment work other than asset-liability modelling We agree that both of these should be excluded. The relevant areas to capital modelling actuaries for which we are invited to comment on whether they should be included or not are: 4.76 a) capital assessment and allocation work performed for purposes other than regulatory compliance; 4.76 b) work performed for one of the parties involved in a merger or acquisition, commutation or capital raising exercise; 4.76 d) actuarial information provided to risk committees (paragraphs 4.48 to 4.50); We don t think that a) should be included, on the grounds that the requirements of TAS I could limit the ability of an actuary to contribute to a general insurers strategic decisions this is not the intention of the TAS. For example, companies frequently require actuarial input when making business planning decisions and timescales are often tight. Current practice for many actuaries would typically involve high-level assessments with potentially significant elements of judgement. While communication of reasonably foreseeable limitations and uncertainty are clearly important in this context (and we believe this to be covered by the generic TASs), a fully documented assessment, with updated data and extensive supporting evidence for all assumptions would often not be possible or useful to the user. b) should be excluded. Again the requirements are so onerous and impractical to apply in a general insurance context that it makes no sense to include in scope. Re. d), we would argue that this should not be included within TAS I now, but should possibly be introduced in the future. Solvency II will clearly impose certain standards on some of an actuary s work in this area, and that will be subject to regulatory approval. However, it remains unclear exactly what work will be performed for Solvency II and otherwise by actuaries within risk management, and how it will be used by risk committees. We suggest that this be excluded from TAS I until 2013, by which time actuaries will be more integrated within firms risk management functions and clear standards can be set for actuarial work in this area. In 5.18, the proposed principle contains the line The dataset should be as up to date as possible. For capital modelling exercises there can be thousands of assumptions, and hence huge volumes of supporting datasets, which in practice are not always kept fully up-to-date unless material changes to the modelling outcomes are anticipated. We recommend that this statement makes reference to the materiality of the assumptions to the outcome. In 6.22, the suggested principle contains the line For work performed at regular intervals, assumptions should be changed only if justified by new data. We believe that in General Insurance, where data is often scarce, actuaries should be encouraged to re-consider assumptions regardless of whether new data is available or not as an important part of quality control. Many assumptions are made with very little supporting data, and the actuary should be able to alter these assumptions using reasoned judgement and subjective factors, rather than only using clear supporting data. While we agree with the rest of the principle, we do not believe the first line should be included. In 6.79, the proposed principle is In estimating insurance liabilities and their variability, explicit allowance should be made for changes in the co-dependencies of risks in scenarios of high stress compared with those of low stress. While this is clearly an important area for actuaries to consider, I d be a little bit concerned about how the proposed wording might be implemented. If some actuaries interpreted this as saying that standard (linear dependency) correlation matrices are not fit for purpose, then some may begin using more advanced statistical techniques without fully understanding them, which might actually promote the perception of actuarial modelling being a black box, which is something we are trying to avoid. I think in some cases the consideration of non-linear tail dependency through stress and scenario testing is appropriate, and therefore I would suggest that changing the wording from explicit allowance should be made for to explicit consideration should be given to would be a good idea.

From the Capital Modelling perspective (as well as Pricing above) we think that there should be two TAS documents covering General Insurance and Life separately. Comments such as the relationship between the selected discount rate and a low risk rate should be explained to the user, and statements about the quality of data (where in GI this is often much more scarce than Life) and ALM are clearly more relevant for Life actuaries than GI. Having both industries covered under one TAS could impose higher standards for GI actuaries in areas that do not necessarily warrant it. h) Data In general the data section is sensible, however: Section 5 (Data) appears to be over focused on reserving. There needs to be more consideration of pricing and capital modelling data requirements. This problem is not helped by the combination of GI and Life standards into the same document as the data required are very different. 5.14: The paragraph "it will be necessary to seek external information" - in GI, this may be difficult, as publically available benchmarks are difficult to get hold of, and we may be quite restricted as to what sources of information we can use. Wording may need to reflect this, perhaps saying "where possible". i) Assumptions The sections seem to be more weighted towards life insurance than general with a heavy emphasis on cash flows and discounting and specific sections on mortality, morbidity, policyholder decisions and management discretion. We believe this issue would be resolved with the separation of the TAS documents. We believe the general principles are all fairly sensible and what one would imagine most actuaries do in practice anyway. The paper asks for views on whether a principle should be included which addresses whether an allowance should be made for the insurance and reserving cycles in the selection of assumptions. In our view this would be very useful since it is probably one area not being addressed as well as it could be by GI actuaries. The paper says that it is excessively prescriptive to include a principle on allowing for latent claims. Because of to the high level of subjectivity surrounding this matter we agree. One of the principles states that an explicit allowance, which could be zero, should be made for events with very low probability but extremely high severity. This implies an extra allowance over and above the normal cat reserves which should be separately identifiable. We think that, due to the nature of the events the paper uses as examples, the selection of reserving assumptions would be almost as subjective as for latent claims which goes back to the point above about the principles being overly prescriptive. j) Models and Calculation We believe it is a sensible approach to leave the choice of methodology to the practitioner whilst requiring appropriate explanation and documentation as stated in 7.2 7.10 states that reports on capital assessment should discuss the liquidity including risk. It is interesting that BAS has picked up on this area perhaps because it is topical. However, from a general insurer s point of view it may not be the most influential of factors e.g. Market risk is likely to be important given that market returns may often drive the profitability of general insurers in a given year. Should Liquidity risk in particular be singled out for special mention? We do not believe so. Principle 7.23 is interesting as it outlines how scenarios should be investigated under which the ability of a firm to meet its obligations to policyholders would be fully impaired. It is understandable what this principle is attempting to address but it is unclear what a company would do with the information resulting from what may be very unrealistic scenarios. Should only feasible scenarios be included?

With regards to 7.26, we agree that there should be a principle requiring the distinction to be made between earned and unearned business. Actuaries are becoming ever more involved with reported numbers, sometimes producing results which go directly in to financial statements, although in may companies this is not the case. As such, there may be a need for governing principles in the area, where actuaries are involved. This isn t this already covered in accounting standards? 7.29 raises an interesting point but is more directed towards good practice. Perhaps, a principle here could be made more general in regards the distinction and treatment of catastrophe and noncatastrophe claims k) Reporting Sections 8.7 8.10 are concerned with monitoring assumptions against experience. In short, assumptions used in previous calculations should be compared with emerging experience when selecting new assumptions. We believe this is logical. Sections 8.14 8.19 consider Best Estimates and Prudent Estimates. In particular, Section 8.19 states that: In the assessment of insurance liabilities, any prudent estimate of liabilities that is presented should be accompanied by a best estimate. The change in the level of prudence from that in the previous such assessment should be explained to users. Section 8.17 discusses giving a meaning of prudence as this would help users. The example given is liabilities will exceed X with probability Y% We agree that a best estimate figure should be given when providing a prudent figure and that the reasons for the change in level of prudence should be explained to users. However, how the meaning of prudence can be defined will obviously vary depending on the particular work involved. For example, the amount/reliability of available data and the time to be spent on the project will affect whether prudence can be defined by a statistical distribution or a more simple method and this in turn will affect the explanation to users. Hence a degree of proportionality needs to be applied. We have not commented on Sections 8.20 8.24 as they are Life focused - considering With-Profits Business and Surrender Values. This is again not useful for general insurers.