Con Keating’s CDC Question time

This note responds to the questions raised by members of the audience of the webinar: Fair, Sufficient and Sustainable Collective Defined Contribution Pensions;  It quotes them anonymously, verbatim, and then offers responses. It is worth bearing in mind that in many circumstances there are more than one options available, and the choice among them is a matter of preferences.

Question Asked

  • Part of the challenge is that at retirement, to convert DC savings to a retirement annuity are very expensive. I have seen costs such as £180 per £1 of monthly income for a guaranteed 10 years retirement payment, to retiree or beneficiary.

It was the cost and complexity of the decisions and strategies needed to convert a savings ‘pot’ to an income in retirement which led to the concept of CDC schemes as a solution to these problems. Bear in mind that this has been referred to as ‘the hardest problem in finance’. Surveys have repeatedly found that the overwhelming majority of employees would like their pension to be an income in retirement; as CDC schemes come into existence, we will have empirical evidence of the true strength of this preference in the real world. It should be recognised that employees are a heterogenous group of different and varying life expectation, wealth, and cognitive ability, and with that their preference for income will differ. This suggests that there should be flexibility in the design of the decumulation forms available to members. For some this may be full drawdown while for others, with for example a bequest motivation, it may be no consumption of the pension pot at all.

The problems of drawdown are double-edged; that one may run out of funds in older age or that one may live frugally, hoard the funds and die with substantial balances.

The extent to which such flexibilities may be offered will depend heavily on their tax treatment.  In the draft regulations the tax and benefit structure rules have come out as:

  1. a) for annual allowance purposes you are money purchase.
  2. b) for lifetime allowance purposes you are DB.
  3. c) for benefit shape purposes you are DB (except that you can reduce benefits if required by the CDC Scheme rules on a uniform basis).

My reading, which may be wrong, is that I do not think a member can flip back to individual money purchase and drawdown except on CDC scheme wind up and conversion of the scheme back to money purchase. (Note that the legislation refers to CDC as collective money purchase) Some clarification of these issues is needed.

Having said all this, the idea that the cost of buying a ten year monthly payable annuity of £1 should be as high as £180, is absurd. The total pension payments are only £120 over the ten-year period; a savings account paying no interest would do 50% better.

  • It’s unclear to me whether the CAR is calculated as the IRR of total assets relative to future cashflows, or is it purely forward looking based on the IRR of future contributions relative to the future benefits those contributions are expected to earn?

The contractual accrual rate (CAR) is the internal rate of return on assets needed to deliver a particular set of projected benefit cashflows. The projected benefits are estimated using standard actuarial techniques, and may contain many assumptions, such as rates of wage and price inflation and member longevity.

When making new awards in any year, the trustees should have regard to their best expectation of the investment returns available to the scheme. The contribution amount or rate may be determined as the present value of the projected liabilities discounted using this rate.

The contractual accrual rate of the scheme is an average of the contractual accrual rates made over time and membership. The table below shows the effect of introducing new awards into a scheme

  Contribution 1.44 1.56 1.71 1.75 1.79
  Pensionable Pay 7.98 8.12 8.32 8.48 8.65
  CAR/E(RoA) 4.75% 4.50% 4.25% 4.25% 4.25%
  Rate (% PP) 18.00% 19.20% 20.52% 20.62% 20.72%
5.50% Scheme CAR 5.47% 5.44% 5.40% 5.36% 5.32%
  Test 1.046        


The scheme CAR was 5.50% before the introduction of new awards, and new awards were made at CARs ranging from 4.75% down to 4.25% reflecting the trustees’ lower expectations of future investment returns. This had the effect of lowering the overall scheme CAR from 5.50% to 5.32%.

Note that the scheme and these awards share a common set of assumptions with respect to the variable which determine the projected values of scheme benefits. Of course, with the passage of time, many of these variables will have been realised and where they differ from the prior assumptions  made will modify the values of projected benefits, and with that the CARs. Obviously for any individual member their CAR will converge over their lifetime to the rate which was necessary to deliver their pension. For a scheme, the situation is slightly different, as it may never expire, but the more mature a scheme is the closer it will be convergent.

So, to answer the question unequivocally: at award the CAR is purely forward-looking but over time it absorbs the realised experience.

The table above also shows one of the ongoing tests required under the draft Regulations. This test requires the ratio of the current future service cost to its five-year average to lie in the range 0.5 – 2.0. It is an appallingly poor test. If we take the example of the table above, to fail the test it would require average contribution rates to have been above 45% or under 7.5%. These are extremes which are so unlikely as to be meaningless.

Note that the CAR is a reflection of the promises made to members, it is used to accrue benefits (liabilities) not to derive present values for assets. The market value of assets is a statement of what has been achieved and the accrued value of liabilities using the CAR is what should have been achieved by the point in time of the valuation.

It may be obvious but for the avoidance of doubt, the CAR of a new award is the required rate of return for that award to be met on time and in full. But contrast, the discount rate of the viability valuation is the long-term expected return on assets held or to be held.

  • If transfers out of the scheme are allowed, how would you go about determining that transfer value?
  • 11) Would the beneficial interest on transfer reflect the member’s CAR?

The transfer value of a member is the net asset value of their beneficial interest in the scheme as a proportion of the overall or total of scheme beneficial interests. The beneficial interest of a member is the present value (or equivalently the accrued value) of their projected pension benefits discounted or accrued (in our preferred model) at the scheme CAR.

This formulation has the benefit that transfers out or in under these terms will not affect the interests of other scheme members.

Though it is possible to conceive of transfer values based on the member’s rather than scheme’s CAR, these would exclude many of the collective advantages of scheme membership.

  • Why will CDC be able to use a viability test based on expected returns when DB is soon no longer to have that option and low risk, low dependency discount rates based on gilts will apply?

We have not yet given up trying to stop the introduction of low risk, low dependency discount rates for DB schemes. It is bad enough that DB valuations already embed excess prudence through the presence of ‘prudent’ biases in both benefit projection assumptions and discount rates.

Using the best expectation of expected returns on assets as a discount rate should produce unbiased estimates of the long-term viability of a scheme. See later for a discussion of the setting of such rates.

  • The questions were subtly different. I’m a fan but doubt employers will go for it

I regret to say that I do not understand what questions the first part of this comment refers to. (Perhaps the author can enlighten me.) As for employers going for it, I think this is a matter of the time and experience necessary for employers to develop confidence. It will also require some extension of the eligible universe. If we see, as we would like, schemes which are independent of any employer, that is, schemes which are chosen by the member, in much the same way as their bank account, then employers do not really need to go for it, their involvement need be no more than the payment of a contribution.

  • Can Con explain to us where his “expected return from assets” at a point of calculation comes from? Who is expecting it?  I have a DC plan and I “hope” for a return of 10% per annum but I am not “expecting” it and I won’t take account of that in advance.  If I get it, I might draw more on my pension but if I don’t, then I won’t.

The draft Regulations specify that “the discount rate must be determined using a central estimate of the estimated future returns on assets held by the scheme or expected to be held in the future”. My interpretation of the term ‘central estimate’ is that this is mathematical expectation or colloquially a best estimate which was defined in an Institute of Actuaries paper as: “It is her/his subjective derivation of the mean of all possible outcomes, taking into account all available information about the business being analysed”. A “hope” is not an expectation.

The answer to who is expecting – it is the scheme trustees, in consultation with their actuary and investment advisors. The key point is that these are unbiased estimates; they are not ‘prudently’ biased.

I would make the point that the problem faced by trustees is one of estimating long term returns not movements in financial markets in the short term. There is a useful analogy; this is a problem of climatology not meteorology. It can be informed, in Bayesian fashion, by the performance of the asset portfolio held.

Note that the expected return on assets used in new award contribution may differ from the expected return on the portfolio assets. For a new award, the question is: what would the market currently offer for new investments; for the scheme, it is a question of what can be expected of the assets held or to be held.

There should also be differences between new issue CARs and the expected return on assets of the coverage. The new award CAR is principally concerned with estimating the benefits awarded to members while the viability valuation needs also to consider provisions for possible future events such as wind-up costs.

It is good to see that the author is following the approach we recommended in the presentation for the payment of bonuses; that these should be paid only once earned. This is why we recommend the use of the supplementary status quo valuation which compares the accrued value of liabilities calculated using the CAR with assets held valued at market prices. This is how have we done so far valuation in contrast to the how do we expect to do of the viability valuation.

  • What is the benefit of a CDC compared to a well diversified SIPP?

I will answer this anecdotally to begin. In late 2007 I helped a close friend to determine the retirement income he could expect given his large highly diversified collection of DC arrangements: it was £89,375 annually using lifetime annuities. One year later, in late 2008, as he was approaching his year-end retirement, it was just £29,825 as the value of his portfolio had been massacred by market developments.

He decided not to annuitise, but to hold off and hope that markets would recover. He could afford to do this, but several world cruises did not happen.

Simply put, the well-diversified SIPP still faces all of the problems of decumulation that CDC was designed to resolve. It is also most unlikely that even a well-diversified SIPP would capture the economies of scale and scope available to a large collective scheme.

  • How does the scheme adjust and achieve fairness if, over time, the investment returns disappoint? For example, many people are discussing investment returns with past long term returns as the start point.   If we have a much lower growth world in future, real rates of return, especially after expenses, might be close to zero on risk assets and below zero on “low risk assets”.  Do you recognise this as a possibility?

A lower growth future with lower real returns, quite possibly negative, is a distinct possibility. The need to bring down the high levels of national indebtedness arising from the pandemic measures seems likely to result in extended financial repression. In addition to this, we have inflationary pressures arising from the changes due to Brexit, the measures needed to counter climate change and  demographic ageing which are likely to prove persistent. In terms of the analysis presented in the webinar, this is an environment in which it is likely that the older members of CDC schemes will be supporting the younger.

If returns disappoint, deficits will arise. If these are not transient, then targeted benefits must be cut. We favour a simple equal proportional lowering of all scheme member’s beneficial interests, for example say a cut of 10% for all. This is clearly fair.

Our risk-sharing rules, which defer cuts operate on the basis that they are fair. For example, suppose we have a scheme which is 50% non-pensioner and 50% pensioner, that the scheme is funded to just 90% of the scheme benefit interest total and that 4% of the scheme beneficial interest is payable. If we pay the 4% fully, then we need to adjust the relative claims such that the non-pensioner members have the same net asset value after the payment that they had before, i.e. 45. As the assets will have declined by the pension payment, the pensioner member claim reduces to 41 of the assets held of 86.

Some have suggested that lowering or raising the level of, say CPI indexation could be used to effect this. We do not favour this approach as it alters the relative claims of scheme members – the younger will suffer proportionately larger cuts than older members. This is actually a form of risk-sharing among members and of course, if markets recover and full indexation is reinstated, it is the younger members who will benefit most, again also in relative terms.

  • Do the trustees chose the CAR and should they be prudent in choosing it.

Trustees set the initial CAR for new awards as noted earlier. Unbiased estimates are necessary in CDC schemes if fairness among members and inter-generationally is to be maintained and the scheme prove sustainable.

10)   In a CDC scheme is any of the contributions allocated to a member?  Isn’t it a collective pot?     What is allocated to each member is the target benefit accrual, which is then subject to adjustment annually?   The contribution rate just determines how much is added to the total pot each year.

A notional allocation of contributions is made to each active member. It is indeed run as a collective pot of assets. We describe the allocation to members as their beneficial interest in the scheme. It is subject to adjustment annually. It is the net asset value of the member’s beneficial interest which represents their claim on the scheme.

For the avoidance of doubt, we favour allocation of the traditional DB type, where the notional amount allocated is based upon their benefit projections discounted (or accruing) at the scheme CAR average.

The contribution rate will usually result in a small modification of the scheme contractual accrual rate – see the earlier table.

12)         Can CDC work just to pay pensions and be funded simply by transfers of existing DC pots?

The answer here is yes but that needs some qualification and explanation. Transfers in and out of existing pots are not problematic in an open ongoing scheme. An open scheme with active members will almost always be able to absorb some level of transfer in, even of pots which are decumulation only. There is scope for risk-sharing among different classes of member. Transfers in are rather sensitive to the pricing on which they are admitted. This is a one-off event unlike the contributions of an active member which occur over multiple years and benefit from that form of averaging. Decumulation only would be an extreme example of this. It would need careful pricing and perhaps rather early adjustments to benefits payable. For decumulation-only risk management approaches using indexation as the control variable have a greatly diminished power.

About henry tapper

Founder of the Pension PlayPen,, partner of Stella, father of Olly . I am the Pension Plowman
This entry was posted in pensions and tagged , , , , , , . Bookmark the permalink.

4 Responses to Con Keating’s CDC Question time

  1. con keating says:

    The responses I delivered to the questions raised during my last CDC webinar have raised a number of further questions and some requests for further clarification. I will produce a further blog covering those when time permits.

    However, as several requested further explanation as to why the expected rates of return on scheme assets and new awards should differ, I will cover that now.

    The term or duration of new awards is usually much longer than the term of the outstanding stock of awards or benefits. In the case of the example shown in the table, the duration of the new awards is approximately 30 years, while the term of the outstanding stock is around 18 years. New awards of course apply only to active scheme members (abstracting from the revaluation of deferreds) while the entire stock includes also both deferreds and pensioners in payment. This means that the sensitivity of these to the factors determining projected benefits differs.

    When considering long-term discount rates or equivalently rates of return, the effect of uncertainty which increases with time is to lower discount rates. There is another mathematical aspect to consider. Suppose we have two possible scenarios for the development of rates: 1) that a rate of 10% will apply and 2) that a rate of 2% will apply and that we consider these equiprobable. Of course, asymptotically the (expected) discount rate prevailing will be the lower rate. For the twenty year, the mean rate is 4.35% while for the 30 year it is 3.88% as we increase convergence to that asymptotic rate.

    I will be doing another Z-Yen webinar in November on the subject of discount rates, an extension of a paper presented at this summer’s SASE conference.

  2. Martin T says:

    “The questions were subtly different. I’m a fan but doubt employers will go for it”
    Context has been lost since I put this in the Q&A box during the event.

    The two questions were put to the audience at the beginning and end of the presentation but the change in responses was erroneously taken as a measure of the event’s success. I thought it was an excellent and informative talk but the answers to the questions were not directly comparable.

    The questions were “Are you a fan of Collective Defined Contribution Pensions?” and “Do you believe that CDC pensions will become an important part of private pension provision?”. My responses were therefore “Yes” and “No” because whilst I think they are an excellent idea, I believe very few employers will want to offer CDC.

  3. con keating says:

    Thank you for the explanation. I had completely missed the point you were making.

  4. henry tapper says:

    I guess more people can see CDC as a part of private provision than are fans of it (you being contrary Martin). We can only guess whether those who see CDC in our pension arrangements were convinced by Con it would happen. Whether they were convinced that was a good thing is another matter.

Leave a Reply