[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Solvency II

iscussions are currently taking place within the EU on the design of a proposed new Solvency II regulatory framework for insurers. The framework includes the approach for assessing the provisions for insurance liabilities, along with the associated minimum capital requirements.

Assessment of provisions
One of the first steps in Solvency II has been to reconsider the methodology for assessing the technical provisions for insurance liabilities. The present methodology is based on the concept that ‘obligations to policyholders should be met as far as can reasonably be foreseen’, and the concept was developed further for life insurers by some more specific principles that were recommended around 15 years ago by the Groupe Consultatif, the European association of actuaries.
However, this concept is interpreted in various different ways across Europe for both life and general insurers. In particular, there is a range of views among both supervisors and firms about the level of prudence that should be included within these provisions.
Accordingly, the European Commission decided, as an initial working hypothesis for Solvency II, to propose a benchmark for this level of prudence, namely that the provisions should be sufficient with a 75% level of confidence to meet the liabilities to policyholders. This in turn is likely to have been influenced by some recent work in Australia, where this was the chosen benchmark.

Best estimates
In order to assist the Commission’s work, CEIOPS, the European committee of insurance (and occupational pension scheme) supervisors recently completed a quantitative impact study (QIS1) in which they invited a range of insurance firms to assess both some ‘best-estimate’ provisions and some provisions based on a range of possible percentiles of the distribution of cashflows, including a 75th percentile that was intended to be broadly equivalent, for most lines of business, to the Commission’s proposed 75% level of confidence. These provisions could then also be compared to the current balance sheet provisions.
For this purpose, life firms were asked to include a value for all future bonuses, including discretionary bonuses, taking account of both obligations to policyholders and policyholder expectations of how management will run the business.
There were of course a number of issues to be considered in order to be able to undertake these calculations, including which risks were to be considered when evaluating the risk margins (ie the difference between the percentile provisions and the best-estimate provisions), and how in practice to make the assessment of the relevant probability distributions for each risk.
The supervisors in most European countries agreed that for the purpose of this first study, the risk margins added to best-estimate provisions, in order to produce the desired level of confidence in the provisions, need only relate to unhedgeable risks, which would include primarily underwriting related risks, persistency, expenses, and the assumed take-up rates for options. Options and guarantees were in principle to be valued on a market-consistent basis.
In practice, most UK life insurance firms adapted their ICAS figures to derive the relevant percentiles for the provisions, while general insurance firms had recourse to a stochastic simulation model to assess the variability in future claim payments.
For UK life firms, the provisions assessed for with-profits policies were, as expected, close to their ‘realistic liabilities’, while the best-estimate provisions for linked policies and annuities were around 5% lower than their current balance sheet provisions.
The study showed that, in most countries, the size of the 75th-percentile risk margins for these unhedgeable risks was quite low, often around 2 to 3% of the best-estimate provisions, though rather higher for protection policies. The more significant factors for the assessment of the life provisions were generally how the firm addressed financial options and guarantees, and the level of future bonuses assumed.
For UK general insurance firms, the 75th-percentile discounted provisions varied on average from around 10% lower than current balance sheet provisions for motor and property, to around 20% lower for liability and reinsurance. The size of risk margins was noticeably higher for reinsurance accepted than for direct business.
As expected, there was considerable variation between countries in the ratio of these discounted provisions to the current balance sheet provisions. However, there was a perhaps surprising level of consistency in the size of the risk margins across European firms. It is not entirely clear to what extent this reflects a common approach to the modelling of general insurance business and to the parameterisation of these models.

Cost-of-capital approach
Firms were also offered the opportunity in QIS1 to suggest their own particular approach for assessing these risk margins. A number of firms proposed a cost-of-capital approach, based on the concept that provisions should be the present value of expected cashflows, plus a risk margin that essentially represents the cost of holding additional capital to cover both the volatility in future possible cashflows and the uncertainty in the assessment of the best estimate of those cashflows. This is a concept that has been taken up recently by the Swiss regulator for their proposed new solvency regime, and is a concept that appears to be gaining support from many insurers, both here and within Europe.
There were relatively few firms that provided figures for QIS1 based on the cost-of-capital approach. However, over the past few months, the European industry has been pressing strongly for the adoption of this approach, and the European Commission have now modified their working hypothesis to include this cost-of-capital approach as an alternative to a 75% level of confidence approach.
If the methodology and algorithms for this cost-of-capital approach can be shown to be both sound and practical, and if enough firms can produce plausible and meaningful figures from this approach in the next quantitative impact study, then it may well be possible to advance this idea further within Europe as an alternative to an approach based on a specified level of confidence in the provisions.

Capital requirements
CEIOPS has now embarked on this second quantitative impact study (QIS2). In this new study, a range of firms across Europe has been invited to provide information about provisions assessed on both a 75th-percentile approach and a cost-of-capital approach. In addition, information is being sought about the potential impact on firms of some possible new capital requirements.
These capital requirements are intended to provide 99.5% confidence that firms will have sufficient assets to cover their liabilities in 12 months’ time, and a TailVar standard is proposed in this context, particularly for general insurance risks. For this purpose, the liabilities need to include the risk margins assessed on one of the two approaches suggested above, in order to ensure that the business could either be transferred to a well-diversified third party or recapitalised by the existing owners.
The calibration for QIS2 was intentionally quite tentative. More work will be needed later to determine the appropriate parameters to match the chosen capital standard.
In contrast to the present scenario-based approach that was introduced as part of the realistic reporting regime for UK with-profits offices, the suggested capital requirements in QIS2 are largely factor-based, as this is seen as being a more practicable approach for European firms in general.
The factors are intended to cover both parameter uncertainty and statistical variability in the provisions and are calculated separately for each risk component:
– financial market risk;
– credit risk;
– underwriting risk (including both persistency and expenses for life business); and
– operational risk.
These components are then combined through a set of correlation matrices (based on assumed tail correlations) in order to allow for some potential diversification benefits between different risks.
However, this does then raise the difficult issue of how to allow for the ability of life firms to reduce bonuses to policyholders in adverse conditions. The CEA, the European industry association, has suggested a so-called ‘K-factor’ approach to resolve this issue, but this is likely to need considerable further development, and the actuarial profession may well have an important role to play in this, even though there is also increasing recognition within Europe of the need to take account of legal issues when considering bonus policy.
For general insurers, the suggested approach for capital requirements in QIS2 bears some resemblance to the enhanced capital requirement (ECR) that was introduced recently in the UK, but with the addition of a component to reflect the potential costs of certain natural catastrophe scenarios. There is also a set of correlation factors that may be taken into account to aggregate the capital requirements for each line of business with reinsurance business being considered as a single class of business for this purpose. Within QIS2, there is also the possibility of making use of firms’ own historical claim ratios, though there is an unresolved issue about how to find suitable means to adjust these ratios for the effect of the underwriting cycle.
More generally, it is envisaged though that firms will be able to utilise their own internal models, subject to supervisory approval, in place of the standard approach for capital requirements. This may resemble in many ways the individual capital adequacy standards (ICAS) assessment currently made by firms here.
There is also likely to be a supervisory review process similar in concept to the individual capital guidance (ICG) regime currently applied for UK firms, but the full details of how this would operate have yet to be decided.

Further developments
The Groupe Consultatif may be invited to provide advice on a number of issues related to the above, including:
– the important starting point of how to assess a ‘best estimate’;
– the valuation of options and guarantees on life policies;
– the allowance to be made for large losses and claims inflation on general insurance policies; and
– how best to allow for the effect of reinsurance programmes.
Of course, all these discussions are still continuing, and there may well be a number of further developments before the Solvency II framework is finalised in time for implementation, expected to be around 2010. However, it is quite likely that Solvency II will result in a number of changes to the current UK regulatory framework, and to the guidance that may be promulgated for the profession by the new Board of Actuarial Standards.