This is a blog in two parts. The first section covers the design of risk-sharing rules which maintain intergenerational fairness among CDC scheme members. The second part marries those rules with the statutory annual viability valuation.
On risk management in CDC schemes
Con Keating and Iain Clacher
It is self-evident that all that may be distributed to CDC scheme members as pensions is fundamentally determined by the contributions made and the investment income earned on these contributions. No form of risk sharing among members can increase the investment receipts of the scheme.
It would be possible to augment the funds available to existing members if new awards were priced to subsidise their pensions. However, this is explicitly unfair to the new members and such transfers are to be avoided if the scheme is to treat all members equitably and be sustainable[i]. All new awards should be priced on their own merits; this should neither subsidise nor be subsidised by the position of existing members.
The ambition of risk-sharing rules is to lower the experienced volatility of the scheme for pensioners. When successful, risk-sharing among members can enhance their comfort and confidence in the scheme. It is, in essence, no more than a form of multi-period smoothing. It exploits the collective nature of the scheme and in particular the differing time horizons and cash needs of members.
The challenge is therefore to do this in a manner which is equitable to all members. Simply paying pensioners in full when in deficit is inequitable to other members and clearly not sustainable.
Decision Criteria and Management Instruments
The primary decision criterion is the Solvency Ratio. This is the ratio of scheme assets to total notional scheme liabilities. The present value of the notional pensions promised to all members, calculated using the contractual accrual rate (CAR) is defined as the scheme beneficial interest. It has a member level analogue, a member’s beneficial interest. The notional target pensions of scheme members are treated as if they are substantial liabilities. In our scheme design, the beneficial interest is the key risk management instrument, and scheme risk-sharing rules operate by variation of members’ beneficial interests. Such variation will alter the contractual accrual rate of members and the scheme.
The contractual accrual rate, which will be described in detail in a later blog, serves as a secondary risk management indicator. The level of experienced returns relative to the contractual accrual rate provides an indication of the long-tern sustainability of the scheme.
The rules which follow may be encoded as smart contracts within the computer system administering the scheme and may be executed automatically. The automatic operation of these rules adds to member confidence by removing the trustees from exercising discretion, and the associated uncertainties of that, which may serve to preference one group over another e.g. pensioners over employees.
At heart, the risk-sharing rules distinguish between pensioners in payment, and active members in accrual and deferred members, with the former being the recipients of support and the latter the enablers of that support.
A Set of Risk-Sharing Rules
These rules define and operate on the basis of the risk-sharing capacity of the scheme. The risk-sharing capacity is directly linked to the proportion of the scheme which consists of non-pensioner member interests. For these members time is not of the essence.
If the scheme is in balance (i.e. asset and liabilities are equal in value), pay pensions in full and everything continues unchanged.
If scheme is surplus, say 10% of scheme ‘liabilities’, pay current promised pensions plus 10% bonus (of the amount being paid)
AND Add a similar proportion to the paid bonus to the beneficial interests of all non-pensioner members.
In other words, the risk management operations are being made on only the current payments and the current beneficial interests of non-pensioner members. The latter operation above will modify the contractual accrual rate applying after the payment. Any modification of beneficial interests will have this effect varying the CAR. The operation increases the relative share or interest of non-pensioner members in the asset pool. These operations, the payment of a bonus to pensioners and the increase of non-pensioner members’ beneficial interests will, ceteris paribus, decrease the level of scheme surplus.
It should be noted that this method of risk management differs from that proposed for Royal Mail. There, the risk management instrument is the rate of in retirement consumer price inflation revaluation. That instrument lowers future pensions by amounts which increase with time; the risk-sharing rules proposed are uniform in time. This inter-temporal difference can carry consequences for intergenerational fairness and equity.
If the scheme is in deficit, say 10%, then, in the absence of operative risk sharing rules, cut the current pension payment and the beneficial interests of all members by 10%, to bring the scheme back into balance.
This will have the effect of lowering the CAR going forward.
The risk-sharing support rules vary the relative interests of non-pensioner and pensioner members. The general principle is that any support delivered to pensioners is matched by an equivalent award to the beneficial interests of non-pensioner members.
Clearly, the payment the full amount of the pension in the year when the scheme is already in deficit will exacerbate that deficit going forward, and of course, the award of an increased beneficial interest to non-pensioner members will also do this.
As we can only be concerned with and offer support to overcome transient volatility of asset values, we need two dimensions to the risk sharing support rules.
To eliminate the prospect of systematic mis-estimation of achievable investment returns, we need first, a time dimension and second, operating in conjunction with this, a magnitude or maximum support amount. The second of these rules eliminates the prospect of a ‘death spiral’ arising as assets are paid out in pensions to the detriment of the residual members.
The experienced return to CAR ratio is informative in the context of mis-estimation of returns and possible ‘death spirals’. It may also be presented, at a cost of greater computational complexity, as the likelihood of achieving the required long-term rate given the experienced return distribution of the scheme’s asset allocation.
The time rule is linked to the magnitude of the deficit. The time to recover before cuts to benefits occur is inverse to the magnitude of the deficit. If we have a 10% deficit, we have ten years in which, provided that the magnitude rule permits the total amount of support, support may be provided, and pensions paid in full. If the deficit is, say, 50% the time to recover is just two years; a 20% deficit would admit a five-year cure period.
If say we start with a 10% deficit in the first year of support, and this increases to say 50%, then the cure period shortens from the nine years (10% deficit with 10 years, minus 1 year off additional operation) to the two years associated with that deficit. If the deficit then recovers to say 20%, the cure period remains that of the 50% deficit. If a deficit persists at the end of these two years, pensions and the beneficial interests of all members are cut by this proportion.
The second element, the magnitude defines the total amount of support available from non-pensioner members to pensioners. It is the cumulative amount which may be offered in support of pension payments within any deficit sequence. It appears that 10% of the beneficial interest of non-pensioner members is sufficient to remove the possibility of cuts in most cases.
It is worth noting that as schemes mature and the proportion of total scheme liabilities attributable to pensioners rises relative to those of non-pensioners, the amounts of support available to pensioners decreases. With no non-pensioner members, there is no available support. At first sight, this might appear inequitable to these pensioner members, but it should be recognised that the beneficial interests of these pensioners have been augmented to the extent that these members supported the pensioners and pensions of earlier generations. This augmentation can be considered as a buffer against the possibility of cuts to levels below the original award target.
The fact that with no non-pensioner members, there is no available support suggests that the use of CDC schemes as decumulation only vehicles should only be undertaken cautiously. Adding pensioners to an existing scheme will diminish the support available to existing pensioner members. This can be accommodated in the terms on which these new liabilities are introduced; the CAR or implicit rate of return on the incoming pensioner assets will be lower than for the existing scheme. The precise mechanism by which this pricing of the decumulation liabilities operates is the subject of another note. Suffice it now to say that the key is most easily seen in the context of a scheme which is in surplus. Here the terms offered to the incoming liabilities must be such as to maintain the level of scheme surplus unchanged after their inclusion.
Many have averred that members will not join or contribute to a scheme which is in deficit. This may be accommodated by offering terms to the new award such that the deficit is unchanged by their entry. Suppose that the scheme is in deficit to say a level of 90%, then the benefits awarded to new awards are augmented so that these are funded only to the level of 90%.
CDC and Statutory Risk Management
The new regulations which enable CDC schemes require an annual viability statement to be prepared. Of necessity, it must be a forward-looking document, and that means that it will in part be reliant on the actuary’s/trustees’ estimates of the future returns of the fund. By contrast the value we ascribe to a member’s beneficial interest in the scheme is based upon the contractual accrual rate, which is a verifiable fact.
While the contractual accrual rate is related to the expected returns on assets in the terms of awards made, it is to be expected that this will vary from that expected at the point of viability valuation. The scheme’s contractual accrual rate may be expected to be below the scheme’s expected return on assets. This would apply also to new awards as provision is necessary for the administrative operational costs. If the contractual accrual rate were above the expected return on assets, it would indicate clearly and unequivocally that the scheme is not sustainable.
Note that if the viability valuation indicates a shortfall, cuts to the immediately payable benefits are required. These may be spread over three years to reduce the prospect of large cuts coming as unpleasant surprises to members, but any cuts made in this way may not be ‘back-ended’ – a uniform rate of application is the maximum acceptable.
The risk-sharing rules we have proposed are based upon the actuality of today’s position – the accrued pensions ‘promised’ to members, their beneficial interest in the scheme and today’s asset values. By contrast, the viability statement will be taking the present value of projected benefits using the expected return on assets while comparing this with today’s assets.
The difference between these is such that the beneficial interest estimate is likely to show a shortfall while the viability statement does not. These beneficial interest relative shortfalls are the trigger for and basis of operation of the risk-sharing rules proposed.
It is, of course, possible that a statutory viability-based intervention could occur during the operation of the forbearance phase of the risk-sharing rules, in which case they will over-ride those rules and require an immediate cut in benefits.
The method of estimating the expected return on assets will be important, and doubtless a matter of discussion between actuary and trustees. Questions such as the amount of any change to it after large changes in the achieved returns will need to be addressed.
The regulations envisage cuts applied to the immediate year and also multi-year cuts. The risk sharing rules proposed operate on the immediate year, and when the forbearance periods or amounts are exceeded, absolute cuts. These latter cuts can be seen simply as cuts applied equitably over the entire term of the scheme. Multi-period cuts of intermediate term are problematic in that they will usually carry consequences of intergenerational unfairness among members.
This approach makes available a number of new metrics of scheme performance – for example, the value for money (or efficiency) of a new award would be the ratio of that award’s CAR to the scheme’s expected return on assets. The ratio of the scheme CAR to the expected return on assets is a measure of the sustainability of the scheme – if greater than one, the scheme is not sustainable.
[i] Fairness or the equitable treatment of all members is a necessary but not sufficient requirement for sustainable CDC schemes.