Con Keating on Pension Valuations

Mr Pension

Here’s one of two articles of Con’s he’s publishing here today; this appeared in Professional Pensions yesterday. Con reckoned he’d draw some flack and he’s been proved right (see the tweets at the end). But as Jonathan Stapleton points out, Con’s not the only one who sees a need for change.


After many years of festering in the background, the pension liability valuation arguments have recently flared up publicly again. A wide range of commentators has expressed disquiet with the status quo. The valuation of liabilities is fundamental to pensions management, and so is the nature of the problem.

Much of the debate, which is very much a dialogue of the deaf, consists of shouting around whether the yield on gilts, or the yield on AA corporate bonds, or the expected return on assets should be used. It should not surprise that this debate has found no resolution, as all are wrong.

It is surprising that discount rates of any kind should appear in the valuation of pension liabilities since they do not figure in determination of the pension payments promised and projected. Longevity, wages and earnings do, but not interest rates or yields. They also do not appear in contributions; they are a simple proportion of current pensionable salary. This has led to the correct description of interest rate hedging as hedging of the measure, the discount rate, not the liability itself.

It is worth examining the treatment of liability valuation in insolvency procedures; the courts have been valuing liabilities for a very long time. These operate from the ground up; the valuation of a liability, which is known as the admitted claim, consists of the principal originally advanced plus the accrued unpaid interest on the obligation. ‘Acceleration’ is not about bringing some projected, or even promised, future to the present but about making the principal previously advanced and unpaid accrued interest as originally promised under the terms of the obligation immediately due[1]. So the holder of a 10% coupon bond issued at par entering insolvency six months after the previous coupon was paid, now has a claim for the amount advanced, £100, plus unpaid but accrued interest of £5 (0.5*£10.00).

No creditor has any right to look to the future and base their claim on what might have been, even if such ‘might-have-beens’ were explicitly promised by the company. If that were possible, all creditors would have created visions of wishful ‘unicorns’ and become billionaires, and, as it implies no meaningful recoveries for the honest creditor, that would have the result that credit would simply be unaffordable or entirely unavailable for just about all. The debates and disputes over the equity risk premium and similar arguments around the use of the expected return on assets should be seen in this light.

Let us consider two zero coupon bonds, which mature at the same time five years from now. One was issued twenty years ago, at a price of £9.23, implying a yield to maturity of 10%. The second was issued ten years ago at £48.10, implying a yield to maturity of 5%. In insolvency, the admitted claim for the first would be £9.23 plus the accrued unpaid interest of £52.86 making a total of £62.09. For the second, the admitted claim consists of £48.10 plus the accrued unpaid interest of £32.25, making a total of £78.35. Here we have two claims which mature on the same date, with values today which differ markedly. These values retain the specific information of the company promises made in support of their issuance. Markets in distressed securities reflect these differences in their pricing. Any single discount rate, no matter how chosen, will return a single value for these two bonds, discarding information.

Pension liabilities may be valued in a similar way. The principal is the contribution made. Together with the projected value of benefits promised, this determines an accrual[2] rate. This is the rate of return which equates contribution with projected benefits. It is unique. It is fixed at time of award in just the same way that the rates of return of the zero coupon bonds were fixed at issuance. It does not gyrate with the animal spirits of any market or any portfolio of assets.

This makes explicit the fact that the cost to the corporate sponsor has two elements; the contribution made and the rate of accrual of that contribution, a fact which is usually lost on scheme members. In this view, the role of the pension fund is to offset or defease the accrual cost to the sponsor employer. The pension fund also serves as security for scheme members.

Pension liabilities may be valued without reference to or use of any external discount rate. Moreover, this valuation will retain all of the information implicit in the contributions and promises made by the sponsor employer. This proposed method reports accrued liabilities at the time of measurement, while current-employed protocols return a discounted present value of liabilities, of which there infinitely many. By any test, the accrual rate is objective; with market-consistency, the objectivity is in the process of selection, not the item selected.

The accrual rate possesses one further, but very important property; it is time-consistent. This means that if it were to be used to discount future benefits, it would return that same value as is calculated by accumulation from the contribution forward. Neither market-consistent yields nor expected asset returns are time consistent. Put another way, rates chosen in these ways will not return the correct value of the original contribution if used in a backwards projection. A consequence of this is that changes in the scheme valuation are unreliable, and form a very poor basis for any decision. Time-consistency is an important property if a company’s accounts are to satisfy their statutory requirement to be “true and fair”; most notably that earnings statements be accurate and reliable.

It is clear that the volatility of liabilities arises principally from its introduction through the discount rate utilised. The accrual rate may change, but that requires revision of the benefits projections, or of the contributions made.

The question which arises immediately is how wrong can these current methods be? I looked at a section of a particular DB scheme. The total liabilities projected amount to £365.29 million. Using a discount rate of 2% the present value of these liabilities amounts to £260.28 million. The scheme has assets, at market value, of £207.44 million, meaning that under the current convention, it is regarded as being 79.7% funded.

Going forward these assets need to earn a return of 3.58% to be sufficient to meet all benefit payment liabilities as projected. The portfolio of investments has achieved a return of 8.21% p.a. historically. The accrual rate implicit in the contributions made and awards of benefits outstanding was, and is 6.07%.

The accumulated or accrued value of the contributions made, the current, accurate value of scheme liabilities is £153.37 million. The level of funding of this, the commitment as originally made, albeit implicitly, is 135.5%. Put another way, the investment portfolio has done extremely well, exceeding the rate of accumulation promised, and with which we were comfortable, by a total of over 35% or £54.067 million.

The discrepancy between the ‘market-consistent’ discount rate based valuation and this rate of accrual method is £106.90 million, or 69.7% of the accurate liability. This is the magnitude of the error introduced by this ‘market-consistent’ discount rate. It is 29.27% of the total liabilities ultimately payable.

The amount which needs to be reported in order to satisfy the statutorily required ‘true and fair’ view of UK companies law is £153.37 million. It is the value which is equitable to other stakeholders, other creditors and shareholders. Clearly, the market-consistent present value of £260.28 million is materially different from this, and would fail any ‘true and fair’ view test.

While the error in this case is that ‘market-consistency’ overstates liabilities, the converse is also possible; in the 1970s when market yields were extremely high, had these conventions applied, the present value of liabilities would have been understated to similar, and sometimes larger, degree. With error on this scale, it is scarcely surprising that occupational defined benefit schemes are widely regarded as unaffordably expensive, and that perverse actions and management strategies should have been undertaken and adopted. Many of these actions have themselves raised the cost of the rump of occupational DB provision.

The current accounting or valuation practices have done more than any other genuine risk factor to destroy the UK occupational DB system; the current methods are simply not fit for purpose. Once we have resolved this, then and only then can we address scheme funding and member security properly.

I expect and await an onslaught of protests; I recognise that the vested interests are substantial.

[1] This abstracts from the ‘stay’ element of an insolvency proceeding which precludes action by a creditor to collect due amounts,

[2] This is quite distinct from the traditional use of the term accrual rate, which refers to the proportion of a pensionable salary payable in retirement, accrual rate of 1.5% of, say, final salary for each year of service.




To prove Con’s point – the protests duly arrived. Thank goodness we have a strong and open press that will not be bullied in the shameless fashion detailed below.





Thank goodness for balanced journalists!



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. Bookmark the permalink.

3 Responses to Con Keating on Pension Valuations

  1. herbertcrumb says:

    Thanks for posting Henry. If I’m being honest, I think this is one of the first times I have been able to read one of Con’s articles and think I have managed to keep up. So brainwashed am I by convention.

    If I understand correctly, each contribution made was and is required to generate a return of 6.07% pa to meet the promised benefit.

    The experience to date has been 8.2%pa which is fantastic. Indeed about 1/3rd better than was required. No doubt this reflects the massive rally in all asset prices (gilts, equities, and the rest). Having made so much hay, the scheme now only needs to generate a return of 3.58% on the assets going forwards. So using your realistic assumptions, about 1/3 in equities and 1/3 in gilts should get us there. If our portfolio delivers more than 3.6% pa, we can reduce the risk and target a lower rate of return.

    Now this is where I get a bit confused. If I took the same projected benefits and discounted them at a rate of 3.6% pa, would I end up with a discounted present value of liabilities equal to the current value of the assets?

    We can at this point start the process of working out whether we can get 3.6, how likely it is to happen, and what happens if we get less than 3.6%.

    Perhaps I missed something but are we not all arriving at the same point: a required rate of return on the assets that can’t be guaranteed and is therefore always at risk of being missed.

    If this were the case, we could put this debate behind us and start focusing on the real challenge at hand. How do we increase the likelihood of achieving the required rate of return to ensure members get their retirement income and how do we put protections in place if something goes wrong.

    Dan Mikulski’s first tweet is unnecessary and a shame as are Con’s final 4 words!

    Perhaps we can be put it down to the frustration of a debate that isn’t serving to advance all of our aims of improving retirement incomes.

  2. These are two excellent and very relevant articles. One has to hope that note is made of the truths of the arguments made and that a prompt review is forthcoming before DB SCHEMES become historic items, as other essential financal mechanisms have.

  3. Ian Mills says:

    I may be misunderstanding Con’s point. However, he seems to be drawing an analogy with what bond holders get paid on insolvency. This isn’t my area of expertise, but it sounds logical that their claim is for the principal plus accrued interest rather than for every penny they would ever have hoped to have received if they held the bond to maturity.

    However, I’m not sure the application of this line of thinking to pension funds holds water. On an insolvency event the trustees of the fund (in the UK at least) would put in a claim for a Section 75 debt, which current regulations require to be calculated on a buy-out basis (i.e. their claim is for an amount of money equal to the premium an insurer would want to charge for a bulk annuity policy to pay all benefits due less the market value of the pension scheme’s assets at that point). So if we’re looking for consistency of treatment on insolvency then surely that is the liability? (Although I must confess I am confused why we’re focusing on what happens on insolvency.)

    Con, what did I miss?

Leave a Reply