CDC Code – Testing Times
This is the fourth article from Con Keating and Iain Clacher on the proposed CDC funding code, the means the Pensions Regulator intends to use to authorise and manage CDC schemes, the first of which is Royal Mail’s. For links to previous articles , scroll to the end of this one.
The first three blogs in this series considered elements of the proposed CDC Code which were not mandated by the PSA 2021 or the Regulations. Simply put, in the opinion of the author these elements are prime examples of regulatory overreach and should not be present. This blog looks at another shortcoming of the proposed Code, that is things which should have been there but aren’t.
The focus is on the two “gateway” tests and the two live running tests contained in the Regulations and the implementation of these proposed within the Code. However, the Code is rather light on detailed guidance for the implementation of these tests. Given the centrality of these tests to both authorisation and ongoing operations – they must be passed, and failure to do so would incur further application fees of £77,000 for each application, – this is not trivial.
Gateway Test No. 1
The first “gateway” test came as a surprise in the Regulations; it considers whether the estimated projected average annual increase in target benefits earned over the first 10 years of the scheme’s operation is at least in line with the expected level of the consumer price index (CPI). However, there is no requirement for a CDC scheme to offer indexation.
Quite what is intended to be achieved by this test is unclear. Beyond some general desire to see that initial promises made be of constant purchasing power for an initial arbitrary period, the purpose is not stated. It is worth noting that this will capture differing proportions of the whole term life-long pension promised; with life expectation fixed at 25 years, at 0% CPI the test captures 40% of undiscounted total pensions promised, at 2%, 34%, and at 4%, 29%. This will however currently skew the asset allocation and expected return of the CDC fund, by discouraging the purchase of gilts, given their negative real returns.
The ten-year period does not eliminate dependence on member mortality as the Regulations specify calculation by reference to “the returns to be achieved on the available assets of the scheme during the remaining lives of the first ten years’ beneficiaries, … and, (c) based on the premise that such projected annual increase is to be applied over the remaining lives of the first ten years’ beneficiaries.”
The Regulations also specify: the “first ten years’ beneficiaries” means— (i) the expected active members of the scheme during the first ten years; and (ii) the expected survivors in relation to the expected active members of the scheme during the first ten years;”
Over the first ten years of a CDC scheme, it is to be expected that some scheme members will retire and become pensioners in payment, and it is to be expected that some active members will change employment and leave the scheme becoming deferred members. Why are these apparently excluded from the test calculation?
The second element (ii) of this is also odd in several regards. First, it assumes that the CDC pension design includes survivor benefits when there is no reason to assume that survivor benefits will be a feature of all scheme designs. Second, are we to exclude the expected survivor beneficiaries of deferred or pensioner members?
These questions and issues were obvious at the time of the publication of the Regulations. It was reasonable to expect the Code to clarify and resolve them; this Code does not do that.
Gateway Test No. 2
The second gateway test is concerned with establishing a minimum value of benefits. The Code is largely silent on this, saying only:
“This test considers whether the value of the benefits expected to be provided to each member based on the first five years of the scheme’s operation is at least equal to the contributions payable to the scheme by the member. This limits the amount of cross-subsidisation between members. This test does not take account of contributions that will be made by the member’s employer on behalf of the member, except those made as a result of a salary sacrifice arrangement.”
The objective here runs counter to the essence of a collective arrangement among members, risk-sharing and mutual insurance, and is problematic in that regard. The Regulations specify the detail required in the test’s calculation.
“(6) The second gateway test is met if the expected value of the rights to benefits of each active member which are expected to accrue under the scheme during the relevant period is at least equal in value to the amount of the contributions expected to be made by or on behalf of the member into the scheme in that period (not including contributions made by or on behalf of the employers other than any contributions made as a result of a salary sacrifice arrangement).
The relevant period is defined as five years from the commencement of operations.”
This may cause problems for perfectly sensible scheme designs such as those with uniform defined contributions and benefits – that was the traditional DB scheme structure. One aspect of current debates around CDC scheme design is whether the contribution rate should be age-related. In fact, the uniform structure offers insurance to younger members against extremely low investment outcomes, while offering higher returns in normal times to older active members. It is worth noting that the calculation must be performed for each member and this raises a crucial but as yet unanswered question – does one failing member result in rejection of authorisation?
There is also a question of interpretation. Can this be done simply by comparing the contribution rate to the product of the inflation adjusted projected salary and the accrual rate, or does it need to be a full valuation of every member’s benefits, which will introduce mortality expectations and discount rate concerns?
As currently set out given the lack of clarity, there is the question of how transfers in from other schemes are to be treated: are these to be considered member contributions subject to the test?
The two live running tests are both concerned with cross-subsidy between scheme members and are seeking to limit this. Of course, this runs counter to the collective nature of CDC which is predicated on risk pooling and risk-sharing.
The Code describes the first live running test as:
“ This test considers whether the value of the benefits expected to be provided to each member in respect of the five-year period beginning with the effective date of the viability certificate is at least equal to the contributions payable to the scheme by the member over that period.”
However, Regulations specify that this test should be applied only to active members. Those Regulations also contain:
“… if the expected value of the rights to benefits …. expected to accrue … is at least equal in value to the amount of the contributions …”
There are several possible interpretations of the term ‘expected value of benefits’ and methods of calculation and as such, this raises many as yet unanswered questions. Does this require a full projection of the members’ expected benefits discounted to some present value or is it sufficient to compare contributions with pension accrual rates? How are transfers-in to be treated? The purpose of a Code is to offer guidance on such matters, but this Code is silent.
The Code is a little better on the second live-running test:
“This test considers the risk of excessive cross-subsidy and is applied as part of the scheme actuary’s annual recertification of viability. This test highlights large differences between the value of benefits that are accruing and the contributions being paid to the scheme that could be perceived as unfair to either new joiners or long-time members. Where it is failed, the cross-subsidy has become too large over a sustained period.
Each year, the scheme actuary must calculate the expected value of target benefits accruing in the year following the effective date of this test, and express it as a percentage of pensionable salary. This is over the active member population as a whole, not on an individual-member basis.
They must then compare the average of the results of this calculation over the latest five years with the rate of contributions paid to the scheme by members and employers. If the value is less than half or more than twice this rate, the test is failed.”
This test compares the current rate of award with the average of awards made in the past five years, as they were made at those times, but this is problematic. For some scheme designs, we should expect the rate of new awards to vary from year-to-year, with the variation of the trustees’ expected return on assets. A decline in expected rates of return from 7% to 5% for an open ongoing scheme would result in failure of this test. The test also fails to take account of the fact that the awards initially made may have been modified, (increased or decreased), by risk management operations subsequent to their award. And we should not forget that the annual nature of viability testing effectively requires annual consideration of such variation.
As with the gateway tests some worked examples would have gone far in removing ambiguities.
Last, we will suggest a further test which should be considered. The annual viability is concerned with the future, but there is a more basic test which we might apply to the performance of any contract: has the scheme performed as promised? Are the scheme’s assets sufficient to meet the promises as were made and are outstanding? This is calculation of the value of liabilities using the scheme’s contractual accrual rate, and the comparison of these liabilities with the market value of the scheme assets.
This metric can be used to inform the decision to alter immediate benefits. For example, if the scheme has delivered performance to date as promised, then current benefits should be paid even if the assets held are not deemed sufficient to satisfy future needs – the scheme can be brought into balance by variation of future benefits. If both metrics indicate a shortfall, pensions and future benefit should be cut immediately.
The test we have proposed here would go far in offering comfort to, and engendering confidence in, scheme members. The assurance that benefits will only be cut if poor investment performance has warranted cuts is much more reassuring than informing members that their pensions will depend on the assumptions of the scheme actuary and trustees as to future returns.
Previous in this series