I published this article last week under a different title and it didn’t get read much . I’m publishing it again because it is very good and because it challenges current received wisdom on DB pensions. Pensions have become needy and as the article explains, for all the wrong reasons.
Con Keating and Iain Clacher
Scheme funding is now the principal risk management tool for DB schemes, a development which has been encouraged by parliament and the regulatory authorities.
It appears that the answer to everything is ever more funding of the pension promise and at the heart of this approach is the risk of sponsor insolvency.
The purpose of funding appears now to differ between secured accrued debt and paying pensions obligations. Historically, for DB pensions, this purpose was to provide security for the promise accrued, or earned and unpaid, up to the date of insolvency.
It has changed to funding being adequate to pay all pensions as they become due or to purchase a contract replacing the as-yet unearned future benefits promised in the case of DB pensions.
This change of purpose significantly raises the level of funding required, which is costly to the sponsor employer. In previous blogs we have commented on the inequitable nature of this latter interpretation.
In this blog we shall investigate the costs that these funding strategies impose on the sponsor employer and conduct a small thought experiment to illustrate the nature and magnitude of the issue.
We abstract from the complexities of a full DB pension scheme, and consider a zero-coupon 15-year bond issued at a yield of 6% p.a. which we can view as a simplified approximation of the costs (of the employer) and benefits (to the employee) of a pension scheme.
The bond matures at par (£100) and the initial subscription is £41.73 (the discounted present value of the par value, where the discount rate is 6%). We choose 6% as the yield on this bond as our analysis of a small number of DB pension schemes has shown the average contractual accrual rates (CAR) of their stock of undischarged awards to lie in the 6%-7% range.
If this rate appears high, given current market yields, it should be remembered that it is the result awards made over a history stretching back as far as the 1960s and perhaps even 1950s. One notable feature of DB schemes is that awards made in the mid and late 1970s may have embedded very high accrual rates – in many cases over 14% at the time of award.
Subsequent experience and revised projection assumptions have lowered this rate materially, to less than 9%. This is principally the effect of far lower than assumed wage and price inflation, countered by increasing longevity.
Returning to our thought experiment, we next consider the position after one year, when the accrued value of the liability is £44.73 on the contracted terms. As above, this figure may be derived as the discounted present value of the ultimate payment (benefit) or the accrued value of the subscription (contribution) using the 6% contractual rate.
Let us also consider the effect of using an externally chosen discount rate, say, the gilt yield for the remaining 14-year term. Assuming the gilt-based discount rate is 1%, then the reported value of this liability is now £87.0.
Ordinarily, these values would be immaterial if no action is based upon them. However, to maintain the comparison with pensions we now introduce a collateral security funding arrangement for the obligation, in the amount of these reported values.
The difference in required funding cash flows is stark. It is immediately obvious that these funding strategies cause this obligation to differ in its effective term or duration, and with that their cost.
The original unsecured bond was 15-years. The secured 6% CAR-based accrued liability has a duration of 9.06 years, and the 1% discount rate variant has a duration of just 2.70 years.
The cost to the sponsor company in the unsecured case, where the obligation is funded at maturity, is 6% p.a. This rises to 10.13% p.a. when funding to the CAR level is required, and 38.07% for the 1% discount rate case.
A sponsoring company might well be prepared to accept a 15-year 6% obligation, and perhaps even a 9.06 year 10.13% obligation, but it is difficult to believe that any company would knowingly sign up for a 15 year 6% bond which could be foreshortened to a 2.70 year obligation at a cost of 39.07%.
This thought experiment illustrates the method by which DB pensions have become ’unaffordable’ even though there has been no material increase in the ultimate amount payable. It is the arbitrary application of exogenously sourced discount rates.
Investment returns and funding strategies
To estimate the cost of these different funding strategies to the sponsor company, we have assumed no investment returns. However, if we assume an investment return, we observe a fuller picture. A return of 6% p.a. (1) is particularly informative. At this rate, there is no cost to the sponsor beyond the initial contribution made at the time of award.
By contrast, an additional contribution of £42.77 (2) is still needed for the 1% discount rate case at the year one valuation. This contribution like all other assets in the fund will need to earn 6% per annum. This means that, in future, it will result in surpluses relative to the 1% discount rate liability valuation (assuming the discount rate remains at 1% for the remainder of the liability and is not decreased by other exogenous factors such as government largesse and QE).
This contribution is effectively an advance to the scheme on which it will earns the subsequent surpluses. If the excess funding (relative to valuation) can be extracted (3) at the time of occurrence (which is unlikely) the advance has an average life of 7.19 years, and 14 years if it cannot be recaptured before the discharge of the pension(s).
The practice of spreading large contributions over several years as a form of ‘deficit repair schedule’ is now common with changes to average schedules published annually in the Purple Book.
While having a schedule may alleviate a sponsor’s immediate liquidity concerns, it does little other than shorten the term of the advance i.e. . and these contributions still all earn the investment return of 6% p.a. in our example.
It may be that a 6% p.a. return on capital is competitive with the long-term investment opportunities available to the sponsor company, but this has the unique feature that, when market discount rates are used, its timing lies outside of the control of the sponsor employer.
As pension funds are often promoted because of their long-term nature, it is worth noting that any contributions to a scheme, beyond the initial contribution, which constitute part the long-term capital of the scheme come about by extinguishing a long-term liability of the company.
There is no net gain in long-term investment. In fact, it may be that there is a reduction in productive investment as the money that comes into the pension scheme is invested in the financial economy i.e. existing assets e.g. equity in the secondary market or credit that is already in issuance.
This thought experiment is intended to illustrate the extreme dependence of the sponsor company’s cost of provision on the level of funding imposed and the shape of this through time. We have kept this to a single valuation, but it is perfectly possible to extend this analysis to include multiple dates, stochastic variation, and adjustments such as ”prudence”,but there would be little by way of new insight to the logic illustrated here.
Taken together, funding is a very inefficient and incomplete solution to the risk management problem of DB schemes, namely sponsor insolvency.