As part of the battle to sort out the interest rate under Solvency II, EIOPA carried out a quantitative impact study on the issue. Known as the Long-Term Guarantee Assessment (LTGA), it required insurers to report their Solvency II balance sheet under a variety of different scenarios, each of which used a slightly different way of calculating the interest rate.
Of most interest to the UK is the issue of whether a “matching adjustment”1 will be included and how it will be calculated. EIOPA have been adamant that firms and commentators should not use the LTGA to draw any conclusions about what the final proposals for the matching adjustment, or any other aspect of the rules, will be. However, given that the LTGA tested 13 different scenarios, one could be forgiven for thinking it would be surprising if the final proposals bore no relation to any of them.
This blog does not attempt to describe all of the proposals that the LTGA tested; these have been well documented elsewhere. However it is interesting to explore some of the implications that the proposals for the matching adjustment might have if they were implemented. Not all of these are obvious from the concept of a matching adjustment.
There is a certain amount of dismay in the industry that the rules that were tested appeared to reintroduce the concept of “admissible assets”. A prescriptive approach to specifying which assets could be valued for solvency purposes was supposed to have been replaced with the “Prudent Person Principle”. However the proposals in the LTGA do place restrictions on assets that can be used to back liabilities if firms are to get any credit from the matching adjustment.
One problem that firms have encountered is that annuities had to be backed with assets that had an investment grade credit rating. This ruled out a large number of swaps which were arranged “over the counter” and hence are not rated. Our experience is that annuity providers make extensive use of over the counter swaps in order to achieve a better matching of assets and liabilities than can be done with bonds alone.
Equity release mortgages were also excluded from the list of possible backing assets. One business model that some companies employ is to run portfolios of equity release mortgages and annuities side by side: the annuity premium is invested in an equity release mortgage thus satisfying the cashflow demands of the latter and achieving a higher rate of return than investing the premium in conventional assets. This business model would be put under considerable strain if a matching adjustment could not be used in reserving.
Reading the LTGA, the impression one gets is that EIOPA intended that the matching adjustment would always apply to life annuities and might be extended to other contracts as well. However, if you drill down into the text, the requirements for the “classic” matching adjustment are that the insurer can only be exposed to longevity risk and not mortality and persistency.
On the surface, this would not preclude annuities. However it is not uncommon for annuities to have a “guarantee period”. For example, if the life dies within 5 years of the annuity starting then the balance of the first 5 years’ payments may be paid to their estate as a lump sum. This is particularly common where a bulk annuity deal has been used to replicate the benefits paid from an occupational pension scheme.
In these circumstances the insurer is exposed to some mortality risk and so is forbidden from applying the matching adjustment. (EIOPA’s argument would be that strict matching is not possible since the insurer may have to make a large and unexpected lump sum payment if the life dies and this cannot be replicated with backing assets.)
The LTGA rules prevented insurers from splitting the obligations under a contract. This meant that insurers could not apply the matching adjustment to the pure annuity cashflows and then hold a separate reserve for the death benefits.
What about general insurance?
A particularly peculiar feature of the LTGA rules is that the “classic” matching adjustment can only be applied to life annuities. Annuities arising from non-life contracts would only be eligible for a matching adjustment if the “extended” approach is eventually adopted (and even then the matching adjustment would be lower for non-life than for life).
A current hot topic in general insurance is the award of periodic payment orders (PPOs) which require the insurer to pay compensation to the insured for as long as they are alive. Hence we expect more and more general insurance contracts to give rise to annuity obligations in the future. Managing these will not get any easier if they are subject to more penal reserving requirements.
It is by no means obvious why life and non-life annuities should be treated differently. One would expect Solvency II to look at the risks underlying the obligations, rather than the type of contract that gave rise to them.
Barnett Waddingham commentary
The original aims of Solvency II were to promote better risk management, harmonise the solvency regimes across Europe and make capital requirements more sensitive to the risk being run. Market consistent liability valuations was not the object and was only ever meant to be a means to achieve these aims. However one could be forgiven for thinking that market consistency has become the main goal of Solvency II. This has led policy makers to place restrictions on the allowance for illiquidity in liability values and it is by no means clear that this promotes good risk management.
At present annuity providers put considerable effort into managing risk by constructing asset portfolios that are well matched with the liabilities and monitoring this hedging frequently. It is therefore surprising that the current strategies do not achieve the optimal capital requirement under Solvency II. This will naturally lead insurers to change their strategy: moving from swaps to bonds and increasing exposure to interest rate changes as a result. This should only happen if we genuinely believe that it is the best way of managing risk. If not then the implication is that the calibration of the capital requirements is wrong.
It is not immediately clear why splitting obligations under a contract, and apply the matching adjustment to the eligible ones, was not allowed. If a contract gives rise to some predictable cashflows then surely it is advisable for the insurer to match them with bonds or swaps and then hold some more liquid assets for the less predictable cashflows?
Regulators should be careful about discouraging insurers from backing annuities with equity release mortgages. This is an innovative business model and could potentially deliver cheaper annuities for consumers. In the current low interest rate environment premiums should not be driven up unnecessarily.
Fundamentally, the LTGA rules are trying to apply a very prescribed set of rules to a myriad of different products and business models across Europe. It is therefore entirely possible that some of the consequences described in this blog are entirely accidental, simply because it was not possible for EIOPA to envisage every possible set of circumstances. The LTGA rules are more detailed than the PRA’s current handbook and have the added complication that they need to be agreed by the European Parliament, European Commission and European Council: whereas the PRA does not need agreement from any legislative body. This goes to the root of why Solvency II has not been finalised yet. A less prescribed and more principles-based approach would surely be preferable.
The issues raised in this blog are not trivial and it is slightly concerning that EIOPA seem to be trying to press ahead with Solvency II implementation before they have been resolved. It would be far better to take the time to resolve these issues now in a way that works for all parties and then, only once this has been achieved, move on to thinking about when Solvency II should be implemented and how this will be done.