Benchmark Rates Liquidity Monitor: Issue 11
We recap the main metrics for total liquidity for interest rate and overnight indexed swaps for December.
The new insurance contracts accounting standard, International Financial Reporting Standard (IFRS) 17 (the Standard), was published in May 2017 and is expected to be implemented in the EU and the UK1 with effect from 1 January 2023. The first IFRS17 balance sheet needed for transition purposes will actually be as at 31 December 2021 in order to derive a comparison set of results for 2022.
The Contractual Service Margin (CSM) is a key component of the new standard that needs to be calculated and amortised over the period of service provision for a group of contracts under the new standard. For new business, the CSM ensures no profit recognition at inception. Determining the CSM approach for new business is challenging, but there is a well-defined approach as set out in the Standard for this. However, for the existing business, a number of different approaches are possible to transition the business to IFRS17 and determine the opening CSM, and generally speaking the Standard is not that prescriptive on these different approaches.
In determining a transition approach for their existing business, firms must initially consider the Full Retrospective Approach (FRA). The FRA requires companies to assume that IFRS17 has always been applied since inception of a contract and roll this forward to the transition date. For many firms, this is impracticable or even impossible, and the Standard recognises this possibility. In such scenarios, one alternative to the FRA is the Modified Retrospective Approach (MRA). The MRA’s objective is to achieve an outcome that is close to the FRA, using reasonable approximations and available information to apply modifications which assist in calculating the opening CSM. However, the MRA still has its own challenges and limitations and in many cases can be almost as challenging as the FRA.
Hence, the final approach in the sequence is the Fair Value Approach (FVA), which is the subject of this paper. Many companies in the UK, the EU and Asia have decided to use the FVA for at least part of their existing business.
This paper is focussed on the options available and market practice across the UK, the EU and Asia. It is less directly applicable to the US where parallel developments in relation to US GAAP are underway.
To determine the opening CSM under the FVA, companies need to calculate the fair value of the business and then deduct the value of the fulfilment cash flows to obtain the opening CSM. The value of the fulfilment cash flows, in IFRS17 terminology, is the combination of the value of the best estimate liabilities using the IFRS17 assumptions, plus the IFRS17 Risk Adjustment (RA) in respect of the nonmarket risks on the business. Hence it is conceptually equivalent to the best estimate liability plus risk margin in Solvency II terms. The inclusion of the RA in this context is an important point to note and (for a given fair value) the inclusion of the RA acts to reduce the opening CSM at the transition date. So a more prudent RA (and there is a choice to be made on the level of prudence of the RA) will mean a lower opening CSM.
The IFRS17 standard refers to another IFRS standard, IFRS13,2 Fair Value Measurement, for determining the fair value. The objective of fair value measurement under IFRS13 is to estimate the price at which an orderly transaction to sell the asset or to transfer the liability would take place between willing market participants at the measurement date under current market conditions. Hence, in general terms, it’s a similar approach, as we often come across it in other circumstances—for example, the Solvency II risk margin (RM) is in principle the amount needed in addition to best estimate liabilities to allow the liabilities to be transferred to a third party. We can regard the fair value of the liabilities as being comparable under IFRS17, with the premium received at initial recognition for the insurance liabilities in question.
The IFRS13 guidance is fine where there is an active and observable market for the liabilities in question. As we will show in this paper, sometimes this is the case, but often there is either no such market, or the available market information is hard in practice to use in a reliable way. In such cases indirect or proxy type methods are needed. IFRS13 does allow this through a dividend discount approach, which is referred to further below.
We should also recognise that commercial considerations come into play, particularly when a variety of methods and choices are available. Companies generally prefer to have a higher opening CSM as that will lead to higher future IFRS17 profits. A higher opening CSM will mean less opening IFRS17 equity or surplus on the IFRS17 balance sheet, but in our experience this is not usually a concern, and the desire for higher future IFRS17 profits is usually the priority. For the FVA, this means determining a higher fair value. Clearly, even if the commercial desire is to have a higher fair value and hence a higher opening CSM, it still has to be properly justified in relation to the relevant standards and in relation to the available market data and market practice.
We can categorise the methods available to insurers to determine the fair value of a block of business as either direct or indirect.
A direct method involves using available market transaction data, and taking the price and/or value implied by these market transactions to determine directly the implied fair value of the liabilities. Essentially, by considering the haircuts applied to own funds in market transactions, we can derive the implied fair value of the liabilities. The Appendix to this paper set outs in more detail how this approach can in theory be applied. The direct method follows closely the definition of fair value set out in IFRS13. Adjustments may be necessary to allow for the difference between the dates of the market transactions utilised and the required date of the fair value determination.
An indirect method involves using one of a number of standard valuation techniques to determine a proxy fair value, or an estimate of the price at which an orderly transaction might take place. This may involve some market-determined parameters, but does not seek to directly make use of market transaction data.
Clearly, the direct approach has to be preferable given the direct link to market transactions and the IFRS13 definition of fair value. However, other than in quite specific circumstances, suitable market transaction data is sparse and is difficult to use. Hence indirect approaches are common. We now explore both methods further.
In the UK an active market exists with respect to annuity business. Pension schemes are actively buying out their liabilities with insurers and these transactions fit closely with the IFRS13 definition of fair value. Several employee benefit consultants are active in this market, acting effectively as brokers, and they are able to use the pension scheme liability cash flows, and the winning insurance tender price, to back-solve for the implied discount rate. This discount rate, which is often quoted as a number of basis points (bps) above or below risk-free rates, includes allowance for all the expenses, cost of capital and profit requirements for the insurer when taking on the liabilities. The employee benefit consultants often publish averages of the results of these calculations in order to monitor and indicate the level of pricing competitiveness in this market.
Additionally, the UK insurers and reinsurers that participate in this market clearly have access to market pricing via the deals which they tender on and win. Hence such insurers have the option of using their own data where they win the case in competitive tender, and at least one UK insurer is using its own such data for IFRS17 fair value purposes.
The discount rate as determined above can be used to discount an insurer’s own annuity cash flows to determine the IFRS17 fair value. However, allowance will have to be made to reflect the actual portfolio that is valued for IFRS17 purposes, as it may not be reflective of the average transactions taking place in the market from which the discount rate has been derived. Adjustments may for example need to be made for:
Leaving aside special market situations such as the UK annuity business, other actual transactions that take place in the life insurance industry can be used as a basis for determining a fair value of liabilities. This involves looking at recent transactions of insurance companies and/or insurance portfolios, and considering them alongside the available Solvency II reported results.
The price paid for a life insurance entity is typically at a discount to the reported available Solvency II Own Funds. This haircut implies that those purchasing insurance entities are not entirely satisfied by the Solvency II valuation of the assets and/or the technical provisions. In other words, purchasers require some additional allowance for risk or for their required returns. From looking at market transactions across the globe, particularly in Europe, this discount or haircut to the own funds applies to both companies which are open and those closed to future new business.
The haircut can then be expressed as an adjustment (i.e., a percentage increase) to the Solvency II best estimate liability (BEL), which we can take as the market’s view of the fair value of the liabilities. The larger the haircut to Solvency II Own Funds, the larger the implied increase in BEL and the larger the implied fair value of the liabilities. Even if some aspects of the haircut relate to asset valuation rather than liability valuation, we can still express the total haircut as a percentage uplift to the liabilities.
This BEL adjustment factor, derived from market transactions, can then be used on an insurer’s own portfolio of business to determine a fair value. However, necessary adjustments would need to be made to ensure that the blocks of business that are being valued fairly for IFRS17 purposes correspond to those underlying the market transactions.
Whilst this approach is attractive in theory, there are some obvious challenges with it, the key one being the suitability of the transactions in question in relation to the blocks of business being valued. For example:
Hence, although this approach has attractions in theory, it is in practice hard to apply and is not (as can be seen from the survey below) in our experience a common method of determining IFRS17 fair value. This has led to the following indirect approaches being developed and used by many companies.
Existing valuation approaches such as European Embedded Value (EEV) and Market Consistent Embedded Value (MCEV) can be used as proxies for a market transaction price. These approaches are a form of a dividend discount approach, which is an approach defined and allowed under IFRS13 for determining fair value.
In using an EEV-based or MCEV based method, consideration needs to be given to whether to apply a reduction to the value to take account of the fact that back-book transactions often take place at less than the EEV or MCEV. This could either be an explicit deduction or (for example) by way of a more prudent discount rate. Any such reductions would be subjective and would lead to higher opening CSMs (via higher implied fair values of liabilities).
Alternatively, for EU countries and countries which have similar solvency systems, the Solvency II balance sheet can be used as a starting point for a calculation of the fair value of liabilities. We can take the BEL and RM under Solvency II and seek to “correct” for the unrealistic features that are used under Solvency II. For example:
We note that these adjustments can go in either direction.
All of these approaches have the disadvantage that they are not using any actual market transaction data. They are thus a view of what market participants could or should be doing as opposed to what they are actually doing. However, these approaches have the advantage that they are capable of being evaluated by the insurer in question and, for insurance groups, can be evaluated consistently across the (global) group.
A further indirect approach is to use a cost of capital technique, similar to the prescribed risk margin calculation under Solvency II.
The combination of the Solvency II risk margin and the Solvency II BEL aims to reflect the amount that an insurer would have to pay to transfer the liabilities to a third party. This is clearly consistent with the definition of fair value.
An advantage of this method is that the IFRS17 RA can be determined using a similar approach but with a lower assumed cost of capital rate. One example is a 3% pa cost of capital rate for the risk adjustment and a 6% cost of capital rate for the fair value, consistent with Solvency II. The opening CSM for an insurer taking this approach is the difference between the cost of capital calculated at these two different rates.
The lower assumed cost of capital rate for the RA might be justified as meeting the IFRS17 requirement for a margin or adjustment in addition to the IFRS17 BEL, but without going as far as the full amount needed to transfer the liabilities elsewhere.
A disadvantage of this approach is that it does not take into account actual market transactions, but instead relies on the fundamental principle (i.e., the technical provisions needed to transfer the liabilities elsewhere) embodied in Solvency II.
In a similar way to extending the cost of capital approach as described above, companies using a confidence level calculation technique to determine the IFRS17 RA can extend this approach to determine fair value. Using a confidence level approach to determine the RA is a popular method given the IFRS17 requirement to disclose the confidence level which underlies the RA. A confidence level approach is also used for the equivalent Margin on Current Estimate (MOCE) in the international Insurance Capital Standard, which is currently being trialled on large international insurance groups.
The Value at Risk (VaR) or confidence level approach models the cash flows such that the resulting provision is at a particular percentile on the distribution. This is similar to how the Solvency Capital Requirement (SCR) is determined (in principle) under the Solvency II regime, where the SCR is calculated at the 99.5th percentile of the distribution.
An insurer using this approach could for example choose a percentile at say 75% for the calculation of the RA, and a higher percentile at say 97% for the fair value. The difference in the provisions at these two confidence levels becomes the opening CSM.
Much as with the cost of capital technique, the choice of percentile parameter for the fair value is arbitrary and not necessarily based on any market transaction information.
Whereas companies using the cost of capital approach described above tend to use the Solvency II 6% cost of capital rate for the fair value, companies using the confidence level approach have not constrained themselves to using the Solvency II 99.5% standard for the fair value, and values less than 99.5% have been used.
The IFRS17 effective date (1 January 2023) is fast approaching, and most companies have now reached a landing on their chosen methods for their back-book transitions. They often include the FVA for at least some of the business. However, work continues on the detailed implementation and on discussions with auditors.
Other than in limited cases (for example the UK annuity market), all the methods of determining fair value involve significant judgement. The range of approaches is wide—and perhaps wider than we have previously seen in other circumstances when life companies need to select a valuation methodology.
During 2021, Milliman has assembled an analysis of the methods being used by life companies to determine fair value for IFRS17 purposes. The results are set out below. The information has been assembled via discussions between Milliman consultants and clients across the UK, the EU and Asia. The results should not be regarded as being reflective of the whole market, but they do serve to illustrate the wide range of approaches being adopted.
Further, it should be noted that most of the companies are using either the FRA or the MRA for significant parts of their business, and that the tables below refer to the use of the FVA for those parts of the back-book where using the FRA or the MRA is not possible.
Of the five UK companies that we engaged with, the direct approach is being utilised mainly for businesses that have a large back-book of annuity business, although one company was additionally planning to use a direct approach for non-annuity business. Adjusted embedded value is the most popular where annuity business is not being considered.
For the five non-UK companies we engaged with, indirect approaches are clearly preferred, focussing on the cost of capital and confidence level methodologies. The lack of any direct approach (in this small sample) may be due to the fact that, unlike for the UK annuity market, active markets for transferring liabilities are uncommon and (as noted above) whole company market transaction data is hard to accurately observe and use.
Based on the foregoing, and adding in some wider thoughts, we can conclude:
Each insurer wishing to use a fair value approach will have considered the available approaches and landed on its preferred option in conjunction with its auditors. In our experience most companies have now agreed the key aspects of their approaches, though we are aware that in some cases work still needs to be done to fill gaps and complete the approach.
Although hard to apply in practice, we believe it is useful to set out in principle how market transactions which involve the sale and purchase of an entire life company entity can be used to derive an implied fair value of liabilities.
As a first step, we can assume that when a price is paid for an entity, we can express the price as being some adjustment to the “Asset – Liability” position reported under Solvency II (or some equivalent approach). We can initially take:
Own Funds (OF) + Subordinated Debt (SD)
This is the net assets of the entity, treating any subordinated debt which counts as regulatory capital as a genuine debt as it does ultimately need to be repaid. In most cases, and certainly for closed books of business, the price paid P is less than the above, so we can determine the implied haircut H as:
H = (OF + SD) – P
Some of the haircut may be due to asset-related adjustments, and some to liability adjustments, mainly normally the latter. But regardless we can express the entirety of H against the liabilities. Hence we can say (using A to denote Assets, RM to denote Risk Margin, and OL to denote Other Liabilities):
H = A – BEL – RM – OL + SD – P
A – BEL(1 + H/BEL) – RM – OL + SD = P
Hence we can back-solve for the uplift in the BEL (i.e., H / BEL), which delivers the observed transaction price P. Haircuts to OF + SD of 15% to 20% are common (even for entities which are open to new business), and the translation of this into values of H will clearly depend on the relative sizes of the other items in the above equation. Where there is a transaction haircut applied to OF + SD, it is effectively saying that (for whatever reason) the allowances for risk and return on capital within the Solvency II (or equivalent) results are insufficient, and the market is demanding an increased allowance. In other words, the fair value of the liabilities is greater than those stated for the purposes of an arm’s length transaction.
For IFRS17 purposes, we would calculate our IFRS17 BEL as normal in the first instance, and then apply an uplift of (1 + H / BEL)% to obtain our fair value.
In theory we could look at a range of appropriate transactions, adjust them to a common date, and take an average value of H/BEL for our uplift. However, as already noted above, the practical challenges in carrying this out are very significant indeed. Most transactions are in respect of overall entities (rather than portfolios of business) and such transactions can allow for factors such as goodwill, future new business, brand value, organisational aspects and (for groups) group diversification. It is not usually clear whether and to what extent these entity factors apply to the different classes of business within the entity, thus making this theoretical approach hard to use in practice.
1UK Endorsement Board. IFRS 17: UKEB [Draft] Endorsement Criteria ssessment. Retrieved 28 November 2021 from https://www.endorsement-board.uk/endorsement-projects/ifrs-17
2IFRS. IFRS 13 Fair Value Measurement. Retrieved 28 November 2021 from https://www.ifrs.org/issued-standards/list-of-standards/ifrs-13-fair-value-measurement/#standard.