2.1. Market Consistency
A market consistent value is often associated with fair value and should be rational and unbiased. Kemp (2009) defines the market consistent value of an asset or liability to mean:
its market value, if it is readily traded on a market at the point in time that the valuation is struck; or
a reasoned best estimate of what its market value would have been had such a market then existed, in all other situations.
Market consistent valuation frameworks rely on financial instruments traded in deep and liquid markets where cash flows can be used to create a replicating portfolio, or market observables can be used as inputs to models.
Solvency II started with the intention of being market consistent. However, throughout the implementation process numerous compromises have been made, and with each compromise it moves a little further away from true market consistency. Some deviations were more justifiable than others as it has been argued that “full” market consistency may be too harsh or not suitable for long-term liabilities. However “partial” market consistency impacts the potential to hedge liabilities. While some of the measures have more theoretical justification, others have been political compromises amongst EU members to ensure the sustainability of long-term business in certain markets.
Another view is that over and above the theoretical benefits of market consistency, Solvency II also wishes to harmonise valuation methodologies across the EU with minimal scope for subjectivity. A market consistent valuation framework is one way to achieve this. In effect, deviations discussed in this section relating to market consistency may also be moves away from harmonisation.
The main areas of concern are the UFR, MA, VA and transitional measures. These all formed a part of the long-term guarantees (LTG) package, with the aim to eliminate “artificial” volatility from the balance sheet of insurers, but also moves Solvency II away from “full” market consistency. In this section, we will discuss the UFR, MA and VA as these are focussed on short-term market fluctuations; in addition, topics such as the extended recoverability period and symmetric adjustment will be touched upon but not explored in detail here.
The transitional measures described in section 2.3 are also a significant deviation from market consistency as they smooth the transition between Solvency I and Solvency II reserves on the in-force business of insurers without taking the full impact of Solvency II from the implementation date.
The risk-free yield curve is based on interest rates swaps observed in deep, liquid and transparent markets; where swaps are not available government bonds are used. However, for long-term liabilities where liquid financial instruments are not available, the curve is extrapolated from a stipulated Last Liquid Point (LLP) to the UFR, which is provided by EIOPA.
For the euro the LLP is year 20 and the extrapolated curve converges to the UFR of 4.2% over a period of 40 years. Anecdotally the 20 year LLP seems to be a political compromise between the European countries. However, in the current low interest rate environment EIOPA has already suggested a lower UFR in just over 1 year into Solvency II. Their suggestion is that, for the euro in 2018, the “calculated UFR” is 3.65% (as opposed to the current 4.2%) and that after applying a maximum year on year change of 15 basis points the “applicable UFR” will be 4.05% (EIOPA, 2017)
With interest rates at unprecedented low levels, many question the suitability of the UFR. There are debates around this area with insurers and regulators alike trying to gauge what the “right” value should be.
The emergence of negative interest rates in countries such as Switzerland and Germany show more acutely the issues surrounding a fixed UFR set before the current low interest rate environment. In general, negative interest rates challenge current economic theory and the calibration of solvency models across a broad spectrum of financial institutions.
Some countries, such as Germany, are more dependent on the value of the UFR than others, due to the nature of the business sold; while other countries, such as the United Kingdom, rely on measures such as the MA. These valuation tools are a part of the LTG package, countries which may object to the UFR but rely on other aspects of the LTG package, are less likely to challenge it.
The UFR is less of an issue in countries where swap markets are deemed to be deep and liquid out to very long maturities.
The MA is a flat addition to the risk-free yield curve, and can be applied to portfolios where the liabilities are fixed cash flows and the assets backing them can be separately identified, organised and managed from the rest of the business activities of the insurer.
MA is based on the portfolio of the insurer, and is essentially a proxy for the illiquidity premium on the assets held to back the liabilities.
As at 1 January 2016, according to data published by EIOPA (2016a), the MA was being used by insurers in Spain (15 insurers) and the UK (23 insurers) with two further national supervisors indicating that it may be used in the future, particularly in respect of new business. The data shows that removing the MA would reduce the SCR ratio by an average of 65% in the United Kingdom and 50% in Spain.
In the United Kingdom it is applied to annuities, however, it might be considered unfair to single out annuities for preferential treatment. That it is restricted to annuities reflects the fact that this was part of the compromise measures agreed as part of Omnibus II.
Basing the value of a fixed cash flow liability on the assets backing it and recognising on day 1 the unearned illiquidity premium is clearly not market consistent. Indeed it is very similar to the Solvency I discount rate used in the United Kingdom but without a prudent margin in the fundamental spread.
This impact can be to reduce the technical provisions at an undertaking level by up to circa 10%, compared to not using the MA, according to data published by EIOPA (2016a).
Investing in this way creates a capital requirement for spread risk although the spread risk has been reduced to capture the fact that the portfolio is exposed to default rather than spread risk. According to data published by EIOPA (2016a), the impact at insurer level can be to reduce the SCR up to a half, although for other insurers it can result in a small increase.
It is possible that this capital requirement can be more than covered by the benefit of the lower BEL coming from the higher discount rate. This appears counter-intuitive as it is the exact opposite of rewarding effective risk management, as it is hard to imagine that the additional yield comes without additional risk other than illiquidity risk.
There are stipulations in the derivation of the fundamental spread on government and non-government exposures; it is also adjusted so that the MA on non-investment grade assets does not exceed that calculated for investment grade assets. Hence additional capital is required when a BBB-rated bond downgrades to BB-rated.
The MA can become negative when spreads are very tight, although it is usually higher than the VA.
The eligibility criteria for the MA are set out in Appendix 5.
In order to use the MA, insurers need to apply for it from their national regulator; however, once approved the insurer cannot choose not to apply it. If the insurer cannot comply with the eligibility criteria or is not able to restore eligibility status, it will be banned from using MA for a period of 2 years. These rules appear to be a deterrent from applying the MA in the first place and also introduce a new risk in respect of the loss of the MA. It is not clear what other benefit this rule serves for the protection of policyholders.
The consequence of a departure from a market consistent framework is the ability to arbitrage rule inconsistencies.
Assets such as equity release mortgages do not meet the rules and can no longer be held directly within MA portfolios. However, many companies have created special legal entities to meet the eligibility criteria. Special purpose vehicles (SPVs) have been used for cash flows from equity release products, commercial real estate mortgages and infrastructure finance (Rule, 2017). In these SPVs securitisation techniques such as credit enhancement features and liquidity facilities are used so as to allow them to issue eligible senior tranches.
Liabilities such as Periodic Payment Order (PPO) annuity payments can, at least in theory, be reinsured on a Retail Price Index basis into a MA compliant portfolio as a means to derive some benefit from the MA.
Arbitrage opportunities such as this have created the growth of an “industry” in structuring portfolios to make them eligible. These increase the cost base of insurers, increase operational risk and add to the complexity of an insurer’s balance sheet.
Finally, the SCR is calculated separately for different MA portfolios and cannot be diversified with the rest of the insurer’s business activities, or between MA portfolios. Therefore, the more MA portfolios, the less diversification benefit that can be taken into account. However, MA can be taken into account for the calculation of the SCR for spread risk, and hence can decrease the SCR. It significantly reduces SCR volatility too.
The VA is a constant addition to the liquid part of the risk-free yield curve, therefore the spread is only applied up until the LLP for each currency and thereafter the adjusted yield curve is extrapolated to the UFR should it also be applicable.
It aims to protect insurers with long-term liabilities from short-term volatility in the market. VA is the spread between the interest rate of the assets in the reference portfolio and the corresponding risk-free rate for each currency, minus the fundamental spread – a haircut for default and downgrade on the qualifying assets. If the market spreads are very wide in a specific country, a country spread can be added to the currency spread.
The reference portfolio is determined by EIOPA and includes assets typically held by insurers to cover their liabilities (BEL).
Unlike MA, VA can be used in conjunction with transitional measures on risk-free rates, and if the VA is used MA cannot be applied. However, in the standard formula it does not respond to SCR shocks for spread risk.
The use of VA should be allowed by national supervisors, but some countries may have an approval process.
As at 1 January 2016, according to data published by EIOPA (2016a), VA was being used by 852 undertakings in 23 countries, the largest take-ups are in France with 217 undertakings. The technical provisions of the French undertakings applying the VA represent 18.1% of the total technical provisions in the European Economic Area (EEA); this is followed by Germany at 10.1% and the United Kingdom at 9.2% and are mainly for the life businesses.
The data show that removing the VA would reduce the SCR ratio by an average of 34% at the EEA level, with the biggest impact in Denmark and Germany at 91% and 85%, respectively; as compared to the United Kingdom where it is only 6%, therefore at the EEA level it is fairly significant especially for certain countries such as Germany.
MA and VA are utilised by different insurers in different countries. Where approvals need to be obtained, processes may not be equally stringent therefore creating an uneven playing field. This reduces the harmonising effects of Solvency II, which is one of the objectives mentioned in section 1.2.
The various adjustments to the risk-free discount rate raise the question of whether it is too conservative for the insurance industry with long-term liabilities. Where specific adjustments were designed for specific products, such as the MA, there could be a case for unification by establishing a fairer adjustment across the whole yield curve and across all products.
This suggests that there is a case, if not for entity specific discount rates, then a discount rate that reflects a good investment grade entity where insolvency can be entertained – AA or A rated discount curves.
The fact that the standard formula treats sovereign bonds in the domestic currency in all member states as risk-free (as stated in section 22.214.171.124) raises questions as not all countries issue bonds of equal credit quality; spreads in certain countries trade at significant spreads above German government bonds. Further, certain supranational, regional and local authority bonds get a similar treatment, irrespective of financial health although there are requirements they need to meet, e.g., around tax raising powers to provide consistency with sovereigns (Solvency II Wire, 2016a).
Solvency II equivalence for other non-Solvency II countries, e.g.: the United States, Bermuda and Japan also introduce non-market consistent valuation methodologies and reduce the impact of harmonisation for insurers across countries outside of the EU. There is already emergence of evidence which show insurers in Solvency II countries reinsuring longevity risk to regions outside of the EU where solvency requirements are not as stringent (Bulley, 2016).
2.1.5. Findings on market consistency
Our survey at last year’s Life Conference in Edinburgh asked whether aspects of Solvency II that were not market consistent should be removed.
Figure 5 shows that most participants were in favour of the deviations from market consistency, and within the working party we have a difference of opinion on the subject.
Figure 5 The aspects of Solvency II that are not market consistent (e.g. matching adjustment (MA), Ultimate Forward Rate (UFR)) should be removed. Conference workshop: Votes 59, average 1.9, SD 1.25.
There may be flaws in following a market consistent approach but in the working party’s view it is through market consistency that effective risk management is rewarded.
In the United Kingdom, annuity business is the most important long-term guaranteed business written, and it has been recognised that default risk rather than spread risk was key and this was accommodated through a higher discount rate linked to the assets’ owned and lower capital requirements.
The front-ending of the illiquidity premium on the assets held by the insurer is a deviation from market consistency, which means that a significant proportion of the UK’s long-term guaranteed business is still regulated on a basis that is close to the Solvency I regime. To ensure effective risk management for MA business it has been necessary to develop strict admissibility and cash flow matching rules, which comes with issues of its own in terms of creating additional costs and complexity.
Further, the impact of market consistency on financial stability has come into question and the effect on long-term guaranteed business. However, overall, it has made significant improvements in encouraging effective risk management in Europe.
2.2. Risk Margin
2.2.1. Cost of capital approach
Under Solvency II, the technical provisions are defined as the sum of the BEL and risk margin.
The risk margin is intended to represent the amount another insurer would require to take on the obligations of the insurer. It is calculated using a cost of capital approach; the insurer must project its SCR in respect of non-hedgeable risks, and apply a prescribed cost of capital charge of 6% p.a. This charge is then discounted at the risk-free rate to determine the risk margin.
The risk margin is relatively large for insurers with significant non-hedgeable risks, and particularly for those that have a long duration, such as longevity.
The underlying concept is that the notional transfer of technical provisions to another insurer would enable that insurer to cover the technical provisions (including the risk margin) and that the run-off of risk margin each year would compensate the insurer for providing capital against non-hedgeable risks. This effectively means that the transfer of technical provisions of an insurer in financial difficulties would be sufficient (on best estimate assumptions) to fully capitalise the transferred liabilities in the future.
2.2.2. Choice of approach: risk margin as prudence
The concept of a risk margin was not an entirely new concept unique to Solvency II.
The Market Consistent Embedded Value Principles (CFO Forum, 2016a), which pre-date Solvency II, include Cost of Non-Hedgeable Risks, which is similar in concept to the Solvency II risk margin, and is presented as an equivalent cost of capital.
When Solvency II was under development, two alternative approaches to defining the risk margin were considered (CEIOPS, 2007, page 12). One was the cost of capital approach described above (which is the Swiss Solvency Test (SST) approach), and the other was a recalculation of liabilities replacing best estimate assumptions (which are by definition 50th percentile) with an alternative point on the probability distribution, e.g., 75th percentile (which is the Australian approach). One of the main factors which swayed the decision in favour of cost of capital is the difficulty of defining assumptions on the basis of a probability distribution.
It appears from the consideration of two alternative approaches that a key driver for the inclusion of the risk margin in Solvency II was the desire to retain some form of prudent margin to the BEL, rather than a specific conviction that cost of capital must be covered.
2.2.3. Projection of SCR
Although the cost of capital approach was selected on grounds of relative simplicity, it requires an annual projection of SCR for the full run-off period of the liabilities, which is anything but straightforward for many insurers.
To calculate SCR accurately at each future duration requires complex projections within complex projections, and this is impractical for many insurers’ models. This difficulty is recognised within EIOPA guidance, which has set out a number of simplified methods. Unfortunately, these methods do not appear to be sufficiently accurate in many cases.
One robust approach to this problem is to define, for each block of business and for each component of SCR, an appropriate “risk driver” which is output by the model, so that it is assumed that that component of SCR moves proportionately to the driver. For example, for the mass lapse component, the risk driver might be the excess of total surrender values over total BEL in each future year. The projected SCR is then determined in each future year by combining the individual elements in the normal way. This approach requires both analysis and understanding of causes of risks and significant testing.
2.2.4. Cost of capital rate
The cost of capital rate of risk free plus 6% p.a. is defined in the European Commission Delegated Regulation (EU) 2015/35, which is the same rate used for the corresponding calculation in the SST.
It was defined on the basis that the insurer to which liabilities are notionally transferred is itself subject to risks and therefore subject to a relatively high cost of capital. In the development of the SST, it was stated that 6% was deemed to be a reasonable estimate for the stressed cost of capital in BBB-rated companies, which broadly corresponds to the definition of the capital requirement in the SST (Federal Office of Private Insurance, 2006).
It may also be relevant to consider whether discounting at the risk-free rate in determining the risk margin is consistent with adopting a high cost of capital rate reflecting risk.
2.2.5. Problems with the risk margin
In practice, the risk margin has been subject to considerable criticism, primarily on the grounds that it is too sensitive to interest rate movements, but also that it is in any case too large.
To illustrate this, in a speech on 21 February 2017, David Rule (Executive Director, Insurance Supervision, Bank of England) stated that total risk margins for major life insurers in the United Kingdom had increased from £30bn at the start of 2016 to about £44bn in September 2016. He also estimated that a 100 bp reduction in interest rates increase risk margins by around 20%. He went on to refer to the “flawed design” of the risk margin, though he pointed out that for existing business the problems are mitigated by the transitional measure on technical provisions (Rule, 2017).
The other issue with the risk margin is that its sensitivity to market risk does not fall within the calculation of the SCR under the standard formula. Consequently hedging the risk margins sensitivity to interest rates results in an increase in the SCR for rates risk.
2.2.6. Potential amendments to the risk margin
An urgent review of the operation of the risk margin is needed.
This could start off by considering two fundamental questions:
∙ Is the purpose of the risk margin to provide coverage for the cost of capital, or is it more a means of standardising a measure of prudence in addition to BEL?
∙ If the former (or if cost of capital is the preferred approach to the latter), is the requirement to cover full future capitalisation of existing business an appropriate element of Solvency II, or does this represent too high (and therefore too expensive) a level of customer protection?
Answers to these questions could then illuminate a review of risk margin.
The following changes could be considered:
126.96.36.199. Changes within the existing overall structure
∙ Reduce the cost of capital rate from 6% p.a.
∙ Align the cost of capital rate more specifically to the insurance sector.
∙ Increase the discount rate used to determine the risk margin.
∙ Make the risk margin more responsive.
∙ For example, use a mechanism to adjust the discount rate similar to the MA or VA.
∙ Use a cost of capital that decreases as the risk-free rate decreases, though the theoretical justification for this is not clear.
∙ Treat longevity risk as hedgeable – justified by the increasing availability of reinsurance
∙ Determine the cost of capital not on the basis of the projected SCR, but of the excess of the projected SCR over the projected risk margin
∙ This would be consistent with the assumption that there would be no risk margin after the transfer of liabilities, i.e., that full capitalisation was not needed.
∙ However, it would introduce circularity into the calculation of the risk margin, requiring an appropriate solution.
∙ Allow SCR calculations to take into account the change in risk margin under the scenario:
∙ This would mitigate the impact of the risk margin by reducing the amount of SCR, which would itself reduce the risk margin further.
∙ However, it would introduce circularity into the calculation of the risk margin, requiring an appropriate solution.
∙ Impose an artificial maximum to risk margin
188.8.131.52. More extensive changes
The possibilities listed above might be considered suitable for consideration by the EU, or alternatively by the UK post-Brexit.