This is a personal and idiosyncratic commentary on the Pensions Institute/Netspar workshop for the exchange of experiences and views of DC plans in the UK and Netherlands. While there was little which was entirely new to me, there was much to make me review my thinking on aspects of scheme design for CDC plans.
Perhaps the most important point made was the caution that the disparity of design evident internationally arises from the fact that these designs arise in the context of a specific history, a particular national social and cultural environment. The process is path dependent, as also must be the degree of change feasible at any time. The relative importance of pillar 2 pensions in overall retirement sufficiency differs greatly due to the relative generosity of state pension and other benefits. These points alone would provide grounds for questioning the wisdom of importing, unchanged, the Dutch model of CDC into the UK.
It became evident that many of the objections to CDC in the UK are based upon transpositions of the Dutch model to the UK. It was surprising that few, if any, of the Dutch participants present were advocates of the UK adopting their model. A repeated theme of many presentations was that a uniform contribution rate associated with a common pension award (accrual) regardless of age was unfair to younger members, that it amounted to subsidy of the old by the young. It is true that the pot of a young saver would be larger at retirement that that of a saver on the verge of retirement – the investment periods differ. The problem with this is, surprisingly, that it is an incomplete analysis.
The first point to note is that the younger member’s pension award is far larger in amount than the older – the older member may have a life expectation of 25 years, while for the younger it might be as much as 33 years – with indexation at 3% pa, that results in the absolute amount paid to the younger member being of the order of 50% higher than the older. The uniform arrangement also captures and resolves far more risk and uncertainty for the younger than the older arising from the length of the pension arrangement, which may be 26 years for the older and 73 for the younger.
When we model, for a uniform arrangement, the relation between the implicit investment return (CAR), the level of contributions and the age of a member for a fixed pension promise, we obtain a far richer picture. This is shown in diagram 1 below. Here we see that when contributions are low and implicit average returns high, then there is subsidy of the old by the young. However, when contributions are high and implicit average returns are low, then it is the old who subsidise the young.
This is truly risk-sharing, a form of mutual insurance, not a one-way subsidy. The extent of this is shown more simply, as a proportion of the younger member’s salary, for this example, in diagram 2. This diagram also shows the average return on contribution (CAR) for a scheme, where the active member population age distribution is uniform. The exact detail of this risk sharing varies with many factors, such as wage progression and the precise form of benefits promised, but the cross-over feature will remain. In addition, any implicit aggregate bias for a member will tend to be eliminated as they age. Put another way they may expect to benefit as they age if there is a net transfer in that direction.
The argument that younger members will not join such an arrangement as it is inequitable is tenuous in extreme when the value of the insurance it provides is considered. When considered as a proportion of the younger member’s salary it is rather modest in magnitude. This is not to say that there will not be winners and losers under such an arrangement, merely that it is not possible for a younger member to know ex-ante whether they will be either a winner or loser or the extent of that.
There are further risk management aspects to a uniform arrangement. The effect of giving and taking support to and from older members is the reduce the variance of the younger member’s expected investment return (CAR) over time while it increases the variance of the older members. The aggregate effect on a scheme depends on many other factors which make it impossible to draw general conclusions.
With all of this said, it should also be recognised that it is a powerful device for introducing ex ante risk-sharing and co-operative solidarity among members into a scheme. While other arrangements of benefits are possible, e.g. unitised deferred annuities, they usually lack this risk-sharing property.
The related issue of intergenerational fairness was raised repeatedly. Strictly speaking the issue is one of overlapping generations and marginal change over the years. This relates to some significant differences between the Dutch arrangements and those being considered herein the UK. As a preamble, it should be borne in mind that most Dutch CDC started life as DB, but when sponsor support proved unenforceable, were rebranded as CDC. They retain many of the characteristics of DB and are regulated in a manner typical of DB.
Membership of the Dutch system is mandatory; it is not in the UK. This raises questions of governance and accountability of the management and it restricts exit as a control mechanism available to members. The Dutch benefits take the form only of a pension income. Transfers or capital withdrawals are not permitted. There is no tracking of a member’s individual “pot” notional or otherwise. Preservation and consolidation are problematic. All of these points have been raised as objections to the introduction of CDC in the UK; in fact, none need apply.
Reverting to the intergenerational fairness question, the issue at the heart of the problems of UK DB has always been one of equitable treatment of members and remains one today. The naked protesters of the 1990s were complaining of a system of priority under which pensioners could receive all, while actives might lose all. The problems today centre on the inequity of the treatment of different classes of member. Thousands of pages of legislation and regulation have failed to address the central issue. No voluntary pension system can be sustainable if it is inequitable among members. This means, in the case of CDC, that a measure of every member’s claims upon the assets of a scheme is necessary.
This is unnecessary in DB arrangements where the pension claim is on the scheme and ultimately the sponsor guarantor. It is sufficient to track the pension entitlements. Transfers out, as are possible in the UK, then require estimation of the present value of those entitlements at the time of transfer. This may be profoundly inequitable among members over time, as the discount rate is arbitrary. Diagram 3 below shows the cash flows, contributions and projected pensions together with the associated capital value (or required funding level) for a full-service pension under the terms of the promise as made.
It is possible with CDC to report both the amount of the pension promise accrued and its current capital value. The aggregate of all, the required funding (equitable interest) for all members may be compared with the value of the actual fund, and surplus or deficit established. The net asset value of the member’s equitable interest may be transferred at any time, without harm to remaining
|Box 1 – The Legislative Position
This description of the current position with respect to legislation and regulation was prompted by a press report on the subject which did not gybe with my understanding of the position or that expounded during the workshop.
The ambition here is to accommodate CDC – the completion of the DC proposition. It is not to implement or enable the complexities of the 2015 “Defined Ambition” Pensions Act. As Royal Mail, with full union support, wish to implement a CDC scheme, the timing is appropriate. Royal Mail are engaged in discussions with the DWP as to exactly what changes to existing legislation or new legislation might need to be introduced. The DWP has indicated that it would see such variations applying more generally, if introduced. It is recognised by all that the parliamentary calendar is pressured, notably by issues surrounding Brexit.
The position of the DWP is that it is in the investigation and analysis phase. The drafting of any instruments would occur later, if it is decided to progress the proposal. Although we are yet to see the parliamentary work and pensions committee’s report on CDC, there many been many indications from that source, including their latest report on “Freedom and Choice” that they will wish to see CDC introduced.
The requirements in fact appear quite modest, and I will divide them here into “necessary” and “nice to have”:
Ensure that a CDC scheme cannot be recharacterized as a DB scheme
Ensure that cutting CDC pensions would not contravene Section 67 of the Pensions Act 1995 or HMRC “authorised payment” rules
Ensure that the collective benefit aspect of a scheme does not conflict with its “money purchase” designation.
Clarify how CDC benefits should be assessed by HMRC
Nice to Have
The DWP have classified these into four categories:
According to some preliminary legal views, it may be that use of the power (contained in the 2011 Pensions Act) to amend the definition of a money purchase scheme contained in the 1995 Pension Schemes Act is sufficient. But an exhaustive analysis of all aspects really is warranted.
members. There is a one to one relation between the funding status and the pensions ultimately payable – a 10% deficit implies that only 90% of promised pensions will be paid.
The division of asset claims according to a DB uniform type arrangement shown here is just one of many possible. By accepting a particular approach, members are agreeing among themselves an equitable, though notional, partitioning of the fund. It is worth repeating that it is clear that transfers may take place at the net asset value of a member’s equitable interest at any time without harm to other members.
The question of selection effects due to freedom and choice where the chronically ill and the extremely wealthy are likely to transfer out was discussed briefly. Obviously, this holds the potential to vary the quality of the longevity projection. Based upon experience with DB schemes, this seems unlikely to harm the remaining members. Indeed, it may even shorten the expected longevity of the fund.
It should also be clear that any risk-sharing arrangements should maintain this equitable balance. The only form of risk-sharing discussed in the workshop was that of variation to indexation, with possible nominal cuts included in that. This has been seen quite widely in the post GFC period in the Netherlands. It is problematic as such cuts are inequitable among members. They are inequitable among pensioners due to the differences in their remaining life expectations. They are inequitable among the pensioner, and active and deferred member classes. Though it is simpler to apply indexation variation to CARE arrangements through revaluation adjustments, it cannot be used in a simple manner to maintain equitable balance among members. Market-based discounting methods will compound this problem.
It is clear that the Dutch thinking is very much rooted in a DB framework and, as with the UK, concerned with deficit repair rather than cash flow and equitable balance management. Indexation and cuts to benefits are explicitly triggered by breach of explicit buffers, the defined funding ratios of 105% and 130% where these are calculated in exactly the manner of defined benefit. The prudence embedded in the technical provision valuation of UK DB may also be interpreted as a buffer. In other words, these risk management operations suffer from being based on an exogenous variable a discount rate, which is a counterfactual to the promise made. The result is arbitrary bias and spurious volatility, which will carry with it excessive intervention and unnecessary cuts to pensions.
The question of what the correct amount of a buffer should be was raised. The answer to this for CDC in the UK is zero. Capital buffers or capital adequacy ratios are commonplace in banking and insurance. They serve to mitigate risks to the classes of creditor (liability holder) junior to equity. As assets are equal to equity and liabilities to creditors, they serve to increase the coverage ratio of assets to liabilities. The buffer functions by protecting one class of stakeholder (creditors) at the expense of another (shareholders). However, with CDC there is only one class, members, and they hold all of the liabilities. Any partitioning is arbitrary and pointless. It amounts to moving assets from one pocket to another in the members’ clothing. The buffer assets, when held, should be run down or called upon when market performance warrants this, and in this they are no different from the other assets.
Methods such as that followed by New Brunswick, which are expressed as likelihoods of being able to pay benefits, and which segment benefit types are simply more complex and less transparent methods of determining buffer levels. With an arrangement such as a 97.5% likelihood of paying a base benefit and 75% for extras, or indexation, the scheme is in effect setting (for plausible investment and other parameters) an aggregate buffer at 25% or more of the fund’s asset value. Indeed, total scheme buffers would be similar in size or larger than the total value of full indexation.
There are several problems with prospective precautionary arrangements such as buffers. They make provision for events which may not occur and this is costly. One of the few things we know about risk-based management is that means that that more things may occur than will. There is also the question of how optimally to distribute these buffers to pensioners, without which questions of intra and intergenerational inequity will certainly arise.
In fact, with CDC schemes the risk-sharing rules substitute for capital buffers. They determine the amount of support available to pensioners from other members. For example, if we consider a scheme which has 60% of the members’ equitable interest attributable to non-pensioner members and 40% to pensioners, then a support allowance of 10% of the total equitable interest amounts to 25% of pensioners interests, the protected class. Incidentally that level of support has proved more than sufficient in the historic record to pay full pensions at all times.
If we consider only the immediate situation we face when a scheme is in deficit, that is to say when the asset portfolio is lower in value than the aggregate members’ equitable interest, and a pension payment is due, then we are facing a current circumstance, a reality, not a possibility or counterfactual.
Other than in the circumstance where scheme assets are persistently and increasingly lower than the members’ equitable interest, it is only when pension payments are due to be made that cuts need to be considered. At this point cuts to the current payment should be considered, but only made if the payment cannot be made be accommodated within the risk-sharing support rules. It is critical that these rules operate, and are seen to operate, equitably among all members. This may be achieved by adjustment of the equitable interests of non-pensioner members whenever support has been afforded to pensioners.
As an aside, it should be obvious that if balance is maintained consistently among the equitable interests of members, intergenerational inequities cannot occur. It also means that the contribution setting process for new members should not be used to repair prior deficits or distribute accumulated surpluses. Indeed, this caution extends to the new contributions of existing members where such actions would be inequitable among other members.
The risk-sharing rules of a scheme need to consider both dimensions of the randomness to which it is exposed. The animal spirits of capital markets require consideration of their persistence and magnitude in a process of forbearance. A simple rule suffices, if a scheme deficit is not self-repairing within a set period then benefits must be cut. To maintain balance among members whenever a cut to pensions is made, a similar and proportionate cut to other members’ equitable interests must also be made. The forbearance rule might be that a 10% deficit should be recovered within ten years, a 20% deficit within five, and a 50% deficit within two years.
This rule needs to be augmented and limited by a constraint on the cumulative amount of support afforded to pensioners. This was touched upon earlier in the discussion of buffers. The total amount of support which may be comfortably afforded depends upon the composition of the scheme membership. To cite a pathological example, support of 10% of the overall equitable interest of a scheme might account for all of the equitable interest of non-pensioners if pensioners constitute 90% of the fund. However, it should also be recognised that support on this scale would double the ongoing equitable interest of the non-pensioner members in the scheme through the adjustment process. When pensions in excess of those warranted by funding are paid, the equitable interests of all other members must be adjusted.
New support capacity is introduced by new members and more widely new contributions of existing active members. It has been said, incorrectly, that new members will not join a scheme in deficit. In fact, new members should join as their contribution is fairly priced and represents value for money (the trustee contribution pricing/award setting responsibility) and will be immediately augmented by the risk-sharing support adjustment. In addition, it should be recognised that the structure of these schemes brings added advantages to members, that membership of the scheme is valuable due to its insurance-like nature.
However, the dependence of support on non-pensioner members does also mean that the admission of members solely for decumulation needs to be considered in the context of the overall composition of the membership. New members will not have participated previously in support operations and clearly would not be doing so in the future.
There is a natural risk metric which may be derived and reported, the unutilised support capacity under scheme rules, which is equivalent in a manner to publication of the amount of a risk buffer. This metric could be used to feed back into the forbearance rule, such that capacity determined the amount and duration of forbearance. This would be in addition to the publication of the solvency ratio analogue, the deficit or surplus which may be reported in both capital value and pension income terms.
The investment of the fund has been widely discussed. It is far longer in horizon than that of an individual DC fund. It should not need to de-risk and accept lower investment returns in the decumulation phase. The horizons of CDC schemes are sufficiently long that empirical findings such as the dominance of investment income in total return become relevant. It may even be optimal to manage schemes on the basis of income and dividend generation. Certainly, there are incentives within a CDC scheme to manage for the long-term rather than the short-term of individual DC funds
There is an important difference in objective between an individual DC fund and a CDC fund. The objective of an individual DC fund should be to maximise the asset value of the fund at all times. This is intrinsically short-term and as such may exclude oblique and Indirect investment strategies. It is also at odds with the desiderata of individual members, who should desire asset values to be low (and investment returns high) during their accumulation phase on to have this desire switch to maximum values during their retirement decumulation. By contrast, the CDC scheme has an explicit target which it should seek to achieve or surpass, and it needs to do this on average. The length of time for that average is of course determined by the risk-sharing capacity of the scheme and its current funding status.
The centrality of scheme rules in CDC suggests that a process of authorisation which considers the proposed rules may be appropriate. With CDC the prime trustee responsibility and discretion concerns the contribution and award pricing process, where judgement and discretion are required. It should be noted that a limitation on the term for which contributions may be fixed, such as the five-year term in Dutch rules, may be avoided by the trustees being able to vary the benefits awarded, lowering or raising these in accordance with their return expectations.
Certainly, the assumptions and expectations formed in this process should be monitored ex post. In general, discretions should avoided: as much as possible should be captured by scheme rules which would automatically apply.
The question of scheme governance arose at several points in the workshop. As member mutual organisations established as trusts, trustees should be elected by the members on the basis of one member one vote. This will bias influence towards younger members (relative to a system based upon the equitable interest on a member, which would bias towards the older wealthier active members), but that is consistent with a “commons” solution.
It is important to give members a “voice”, which suggests that a minority of members, say 20%, should be able to call extraordinary general meetings for discussion and direction of trustees on matters of governance, including the removal or appointment of trustees. All polling on motions considered by general meetings should be on the basis of the overall membership, not a show of hands at the meeting. Changes to scheme rules should require a super-majority of members, say 75%, and even then, may be subject to regulatory confirmation.
The ambition of the Friends of CDC is quite limited. It is completion of the DC pension proposition. The addition of decumulation and retirement income options to what is currently only a tax-advantaged savings scheme. It is not implementation of the entirety of the 2015 “Defined Ambition” pensions Act. The current position with respect to legislative and regulatory development is covered in Box1.
This note has offered a particular view, design and management method, and it has pointed out some of the difficulties arising with other arrangements. This is not to say that any of these problems could not be overcome through sound management by men of good will. There is, or at least may be, a systemic strength in diversity.