Con Keating explains why John Ralfe is wrong on CDC


This is a commentary on an article which appeared in FT Alphaville, which was sent to me by a friend in Bermuda. It is a description by John Ralfe of a lecture given by him.

The lecture was hosted by Imperial College, and chaired by David Miles of Imperial College but that in no way implies that they agreed with his conclusions.

I attended that lecture, virtually, and my overwhelming memories of it are of the number of questions which Ralfe failed to answer. It was also notable in that many of his generic criticisms of CDC were drawn from and reliant upon the specific model proposed for Royal Mail.

For ease of distinction in reading, I have italicized the original.

Those of you who are employed by the private sector might have noticed — perhaps with some dismay — that over the last two decades, defined benefit (DB) pension schemes, in which an employer guarantees a pension based on employees’ salary and years’ [Sic] of scheme membership, have become about as common as hen’s teeth.

The hyperbole is striking. DB schemes are not rare to the point of non-existence. DB schemes in the public sector have not closed to any marked extent. Private sector DB schemes have closed to a considerable extent but still over 50% by membership are open to future accrual. We might debate the reasons for the closure of DB schemes and in that debate, I would take the side arguing that this trend was the result of incorrect accounting, misguided, if well-intentioned regulation and an over-zealous Pensions Regulator for whom the elimination of all DB would be the elimination of all risk from these schemes.

Such schemes have been replaced by defined contribution (DC) schemes in which there is no guaranteed pension amount — everyone has their individual pot and must manage their own longevity and investment risk. These DC schemes, therefore, can be complex to manage, easy to get wrong and, unlike DB schemes, can lead to people running out of money before they die.

So [Sic] a kind of middle-ground pension scheme known as “collective defined contribution” (CDC) has been touted in recent years as a way of bridging the gap between the old, often more generous, DB schemes and the new DC schemes. But are CDC pensions really the answer?

“Tout” is an interesting choice of verb, with its overtones of spivery and disreputable dealings, when recommend, promote or support would be apposite.

This newspaper certainly believes they have potential, suggesting in a recent editorial that CDC could be a way of giving the young a fairer deal. Royal Mail also thinks CDC is the answer. As part of seeing off threatened strikes in 2018, the company agreed with the Communication Workers Union to replace DB and DC schemes with CDC pensions for its 140,000 staff, and spent months lobbying for new legislation, which has just become law.

It should be noted that the threat of strikes arose from the threat of closure of the Royal Mail DB scheme and its replacement by DC.

But it’s worth noting that UK CDC is being introduced at the same time that the Netherlands — which set up CDC in an attempt to rescue DB, and is the CDC poster-child — is now abandoning it and moving to DC.

This is a travesty. There are only a handful of schemes, of which I designed one, in the Netherlands which were created as CDC. The majority of Dutch schemes were considered DB and marketed and promoted as such to scheme members. It was only when it was recognised that there was no enforceable debt on the employer that these schemes were rebranded as CDC. The significance of this change was poorly communicated to members. There was no attempt to rescue DB, these schemes could be and were obliged by the Dutch Regulator to cut benefits when in deficit.

CDC works by setting a “target pension” amount for members each year, based on member and employer contributions. Target pensions, including pensions in payment, are not guaranteed, but move up or down every year, in line with the value of assets.

Pensions in payment do not need to rise or fall with the value of scheme assets if inter-member risk sharing support is in operation. It is possible to eliminate most and in some circumstances all of the transient volatility of financial markets.

Fans of these schemes claim that for the same contribution, average CDC pensions are much higher than DC, because CDC can capture the expected long-term return from holding equities, but with a lower risk of pension cuts, thanks to “inter-generational risk-sharing”.

The claim made is that for these schemes there is no need to de-risk the portfolio in the manner that is usual for individual DC as the individual approaches retirement. Typically this involves the sale of growth assets and their replacement with lower-yielding bonds. The higher returns this affords the scheme will mathematically reduce the likelihood of cuts to pensions in payment. The support offered under inter-member risk sharing arrangements can further reduce the likelihood of cuts by eliminating the short-term transient volatility of financial markets.

What is “inter-generational risk sharing” and is it fair?

If “inter-generational risk-sharing” sounds woolly, that’s because it is. Explanations of how it produces higher CDC pensions are long on arm-waving, and short on practical nuts-and-bolts.

The term “intergenerational” that Ralfe favours is somewhat misleading, risk-sharing can only take place among the members of a scheme, among the cohorts of a scheme. Provided that is fairly organised, it will also be inter-generationally fair.

Iain Clacher and I have published on numerous occasions papers which detail the way in which risk-sharing support mat be organised equitably among members. We do not claim that higher returns arise from the risk-sharing support among the cohorts of a scheme but rather that this arises because of the longer investment horizon and term, as well as the stability of the asset allocation of the scheme.

I will be presenting our work publicly again in an FS Club webinar and a UCL lecture in late August and early September.

If CDC and DC asset allocation are identical then, by definition, investment returns — and annual changes to target pensions — must also be identical.

It is true that if asset allocations are identical investment returns are identical if costs also are. However, the asset allocation of individual DC will de-risk as retirement approaches, to avoid exposure to the volatility of financial markets in retirement, which comes at the cost of accepting lower returns.

Target pensions are not linked to the value of assets. The value of assets determine what may be paid as a pension at a point in time in retirement, in the absence of operative risk-sharing support among members. As we noted earlier, that support which is defined in amount and duration can eliminate most or all of the volatility- the deviations from trend of financial markets.

Take a CDC plan with 10,000 members and £500m assets. If assets change — up or down — by 20 per cent over a year, then target pensions also change 20 per cent, to restore balance.

This is also erroneous. As noted above, target pensions do not vary in this manner. A particular pension payment might decline by 20% if the scheme was previously in exact balance and there were no operative risk-sharing support arrangements.

Suppose instead we have 10,000 individuals each with a £50,000 DC pot, or £500m in total. If the value of their DC pot changes by 20 per cent over a year, then their “target pension” — what they can expect to draw as a pension — also changes by 20 per cent, exactly like CDC.

This is a repetition of the errors above.

Can CDC schemes take more risk — namely by holding more equities than DC schemes — because they have a longer “time-horizon” than any individual saving for a pension? Royal Mail thinks so: astonishingly, its CDC plan will hold 100 per cent equities.

This is another repetition of earlier errors. The collective nature of CDC affords the scheme the ability to tolerate higher volatility than any individual.

But since CDC target pensions are adjusted annually, their “time-horizon” is not actually “long term” at all — it is just one year.

Target pensions are not adjusted annually; they are as awarded unless subsequently altered by changes in experience of, or the assumptions determining the factors (eg longevity) underlying the projected pension payments. The targets may also be modified by the operations of risk management and risk sharing rules. This does not make them short-term in nature. The target pensions could be updated continuously in real time, but that would have no effect on the asset allocation of the scheme.

 If a CDC plan, say, doubles its equity exposure, it increases both annual expected equity outperformance and the risk of annual underperformance, and therefore the risk of cuts to annual target pensions, including pensions in payment.

The risk of cuts to targeted pensions does not increase in this manner. Cuts to targeted pensions would only increase in this manner if there were no operative risk-sharing support arrangements.

And anyone wanting this higher risk/return can get it directly in their own DC pot.

This simply is not true. If an individual were to adopt this strategy, they would need to have a risk-tolerance, or preparedness to accept cuts which is massively higher than that of an individual in a collective arrangement.

The Royal Mail’s claims to ensure fairness to all members. CDC must be designed so each age cohort receives a pension equal to member and employer contributions, plus investment returns — no more, no less. If a cohort receives more, it is being paid from other members’ contributions.

At last, something which is true, at least it would be if those are experienced (risk-adjusted) returns.

And this is the Royal Mail CDC’s secret — the oldest members receive more than their contributions plus returns, with the extra money paid by younger members, who end up receiving less than the the [Sic]  total of their contributions plus returns.

This is a criticism of the uniform rate of award that is standard in final salary and CARE DB schemes. It fails to recognise that the risk arising from younger members is massively higher than that of the old and this is borne by the scheme.

Take two members, both earning £30,000, both in the scheme for five years — so the same cash contribution — and the same target pension of 1/80th salary increased at CPI plus 1 per cent each year. One joined age 62 and is now 67, starting to draw their pension; the other joined at 22, is 27, and is now leaving Royal Mail, still 40 years away from a pension. Both of them has paid into the pension scheme for five years.

It is more than a little unusual that a 22 year-old and a 62 year-old should have the same starting salary.

If Royal Mail achieves its assumed investment return every single year the oldster’s contributions plus investment returns run out by age 82.5 years, about five years before the average life expectancy assumed by Royal Mail.

I cannot replicate or verify this statement as there is insufficient information given.

What about the 27-year-old? Once they reach their retirement age, the five years’ worth of contributions plus assumed returns for the previous 40 years, is worth so much that it can then generate more than the annual target pension amount in investment returns alone.

Again, there is insufficient information to replicate or verify this. However, it is worth noting that with 2% annual revaluation, 2% annual inflation and 1.5 years per decade increase in longevity, the expected pension of the younger member is almost four times as much in undiscounted lifetime total as the pension of the 67 year-old.

The extra five years of pension paid to the oldster is pinched from the youngster, whose pension is a fraction — about one-fifth, using the Royal Mail’s assumed returns — of their contributions plus assumed returns.

The comment immediately preceding this extract made that point that the total amount of pension expected to be paid to the younger member is approximately 4 times that of the older. When this is considered, the five years of support afforded to the older member actually represent close to 7.5% of the younger members target pension, not one fifth.

It is notable that risk and uncertainty is absent from John Ralfe’s analysis. The older member faces uncertainty and risk spanning the 22 years of retirement, but the younger member faces far more, spanning some 72 years. If we were to consider this uncertainty to consist solely of investment risk uniformly distributed over time (which are most unrealistic assumptions in the real world) the standard (square root of time) scaling of financial theory would apply, and we would observe that the younger member brings almost twice as much risk as the older to the scheme. Of course, in reality the risk associated with the younger member is far wider and higher than this.

In terms of fairness of this uniform age-independent award structure, the relative uncertainty faced by the younger constitutes a ‘veil of ignorance’ behind which the contract is fair in Rawlsian terms.( and in other pieces John Ralfe has been known to cite this concept).

Of course, the uniform age-independent award structure has been a staple of DB schemes throughout their lifetime, and no challenge such as Ralfe proposes has reached the courts. One of the real advantages of the uniform age-independent structure is that it introduces binding risk-sharing among members from the inception of their membership. It initiates all into the collective.

It is also interesting that the illustration chosen is rather less than a purely collective scheme – indeed it is individual for rather longer than it is collective. The scheme envisaged is individual for 70% of its expected lifetime.


The new CDC legislation is weak, allowing Royal Mail to design a CDC structure that is grotesquely unfair to younger members. Royal Mail’s “inter-generational risk-sharing” is really “inter-generational pocket-picking”.

John Maynard Keynes once wrote: “Words ought to be a little wild, for they are the assault of thoughts on the unthinking.”, but the use of “grotesquely” and “pocket-picking” here is misrepresentation. “Grotesquely” for all of the reasons already covered, and “pocket-picking” as this is really agreed inter-cohort support.


A fair CDC design requires four conditions:

An interesting assertion – unsupported by evidence or argument.

First, all CDC plans should have to use long-term market interest rates to value target pensions — as in the Netherlands — not leave it up to each plan to decide.

No, the discount rate utilised in valuation should be the contractual accrual rate, the rate implicit and embedded in the terms of award. As we have written about the contractual accrual rate in many other places, we will not labour the point here.

It is notable that the setting of the long-term rate used in the Netherlands has proved a disputatious process.

Second, annual target pensions should be age-related — higher for younger members, to reflect “the time value of money”, rather than the same for all ages, as at Royal Mail.

No, this is simply a reiteration of the flawed earlier arguments.

Third, target pensions should be in real-inflation-linked, not nominal terms, as at Royal Mail, to discourage “money illusion”, and so that members can understand their real-inflation-linked target pensions.

No. This is a matter of member communications, it really is not a question of scheme design.

Fourth, companies should also have to offer DC with the same company contribution, alongside CDC, so staff can choose which scheme to opt for. Royal Mail is closing its DC pension, forcing staff into CDC.

This is not a matter of CDC scheme design, though it would have cost implications for the sponsoring employer and the ongoing CDC scheme, so it would not be a harmless action.

A fair CDC plan, meeting these conditions, would, of course, have the same overall annual returns as DC, with an identical asset allocation.

This is yet another reiteration of an earlier (debunked) argument. With identical asset allocations the economies of scale of the collective would mean that per member costs would be lower than for the individual.

The more important point is that of the four conditions proposed only two relate to CDC scheme design and the remaining two are inappropriate.

And for virtually all members, CDC has the huge disadvantage of sub-optimal asset allocation versus DC, because asset allocation is identical for all ages. Unlike DC, members cannot change their asset allocation to suit their personal risk preferences; most 30-year-olds, with a long time to drawing a pension, would want to hold more equities than, say, a 70-year-old already drawing a pension.

This is amusing given the earlier implicit criticism of Royal Mail for proposing 100% equity – “cakeism” perhaps. In fact, the issue for a scheme member is the total return achieved in their lifetime, not the asset allocation relative to their age or risk preferences at a point in time. The collective nature of the CDC scheme with its inter-cohort actually means that, in addition to lower costs, members would face lower risk or variability of the scheme’s assets than an individual varying their asset allocation in some life cycle manner.

CDC does have the genuine advantage over DC of pooling longevity risk, and if you live longer than average you are less likely to be squeezed. But longevity insurance comes at a cost — those dying “early” pay for those dying “late”, and many people would choose DC, allowing then [Sic] to leave their unused pension to family.

This argument is as old as the hills. It, of course, fails to consider the other side of this coin. If the member lives longer than expected, then the DC saver will have no pension and will have to fall back on the family (or perhaps the state) for support. In recent decades of course longevity growth has been the dominant concern. Oh, but wait a minute, that is just collective risk-sharing among a different group.

It is difficult to see how CDC does anything to solve the UK’s long-term pension problems. Meanwhile, the Royal Mail version — the only CDC game in town — looks like a con trick by the company and the union that will benefit only the older members, with government collaboration in the form of its half-baked CDC legislation.

Royal Mail is the only scheme to have declared its intention to offer CDC and the Department of Work and Pensions specifically accommodated that desire. There are other potential schemes sitting in the wings but without the detail of the as-yet unpublished regulations, they could not commit unconditionally in similar fashion. There are also a number of entities which would like to introduce multi-employer CDC, which would require an extension of the current legislation, which restricts CDC to single employer.

It is only with wilful blindness that an observer could fail to see any of the solutions offered by CDC to the UK’s long-term pension problems.

It is sad that this rebuttal of the Ralfe critique should run to over twice the length of the original article, but it seemed appropriate. For those who might say that I have not advanced a superior model in the rebuttal, let me say that we (Iain Clacher and I) have written a short description of CDC at the request of an IFoA group which we are happy to share. I will also be doing two (free) public presentations outlining our preferred scheme design in late August and early September.





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.

3 Responses to Con Keating explains why John Ralfe is wrong on CDC

  1. Derek Benstead says:

    Let’s pick a few things out for remarks. Picking these things out doesn’t mean there isn’t anything else to comment on!

    “Third, target pensions should be in real-inflation-linked, not nominal terms, as at Royal Mail, to discourage “money illusion”, and so that members can understand their real-inflation-linked target pensions.”

    Royal Mail CDC target pensions are real-inflation-linked. The initial target annual increase is CPI+1%. Nothing is guaranteed in CDC – neither the initial 1/80 of pay award nor the rate of annual increase. Whether the annual increases average out at more or less than CPI will depend on the scheme’s experience over the years.

    “Second, annual target pensions should be age-related — higher for younger members, to reflect “the time value of money”, rather than the same for all ages, as at Royal Mail.”

    This is a feasible CDC scheme design. Were a master trust or NEST to set up a CDC scheme for anyone to join (whether there is an employer’s contribution or not) this is what they would need to do. Then, as John says, every contribution would be exchanged for a CDC pension credit of equal value. Which means younger members would get a bigger CDC pension for £100 contribution than older members for £100 contribution.

    This is not what Royal Mail have done. The Royal Mail CDC design uses the even rate of benefit accrual for all, long established as a fair minded approach in the world of DB. This only works if there is an employer’s contribution, so everyone regardless of age gets a pension worth more than their own contribution.

    The RM CDC design works on the premise that the scheme as a whole receives the 15.2% contribution rate, and the scheme allocates the same rate of pension accrual (1/80 of pay) to all. It doesn’t work on the premise that each individual member receives 15.2% of pay, although that is a reasonable alternative design.

    “It is difficult to see how CDC does anything to solve the UK’s long-term pension problems.”

    DB run in a highly risk averse way is too expensive and is in grave decline. Individual DC is not a pension at all, unless and until an annuity is purchased. Collective DC does solve the problem of providing an income-for-life in way which is affordable for the contributors with fluctuating but tolerably reliable benefit outcomes.

  2. John Mather says:

    The fundamental flaw in DB is that the potential beneficiary has a greater life expectancy than the institution making and underwriting the promise. i.e. the Company will be dead before the pensioner the liability damages the value of the original host. The PPF is the destination and that is underwritten by the current DB cohort, all members have the same flaw in their construction.

    Madness is doing the same thing repeatedly and expecting a different result. DB exists to justify Public Pensions. Our children will not forgive us for the £ trillions of obligations

  3. Eugen N says:

    I certainly agree with one of the comments: “ Madness is doing the same thing repeatedly and expecting a different result.” we had with profits and with-profits annuities. The with-profits annuities have not paid better than a guaranteed pension annuity. Both Prudential and the other provider stopped selling them because there was no demand anymore.

    Another thing which is stupid is to focus on volatility and not on total return. It is one thing if the scheme gets 6% per annum over the next 30 years or 8% per annum on the same asset allocation. The value of the pensions paid would be total different, they could be 50% higher on 8% per annum than on 6% per annum. We are scaring people with volatility when volatility was not a problem in the last 40 years. In the years of fiat money, asset prices will get protected by Central Banks. It is the total return which could suffer in the end because of the lower GDP growth that may result.

    I work with people who retire, with CDC you try to throw all people in the same pot, when probably 70% will transfer out. The way people are working and retiring is changing. It is not anymore at age 65. People have different risk capacity, some have more other assets than others and like to withdraw in the way it suits them, including to minimise taxation, make use of personal allowances etc .

    In conclusion, we were there before, with with-profits, the Netherlands were there as well with schemes which are CDC, not DB even if they are named DB-lite.. The Swiss have something similar as well, but at least they take no risk and award 0.3% per annum return and 5% at age 60 as conversion rate.

    Let’s not get carries away, with CDC, the only advantage is for people who live longer they get the mortality gains of people who live shorter. I do not think that in the 21 century this “gain” on other people’s death is anymore acceptable, this is why enhanced and impaired pension annuity came to the market to offer a fair deal to these people.

Leave a Reply