Iain Clacher and Con Keating
The Pension Schemes Bill introduces some new obligations for DB scheme trustees in addition to those already existing in Part 3 of the Pensions Act 2004, which is concerned with scheme funding.
All existing obligations are enduring and the new obligation is short and simple: (1) The trustees or managers must determine, and from time to time review and if necessary, revise a strategy for ensuring that pensions and other benefits under the scheme can be provided over the long term. This is referred to in this Part as a “funding and investment strategy”.
The expression “Long Term Objective” and all that goes with that. ”Fast Track”,”Bespoke”, etc., are inventions of The Pensions Regulator.
The Bill expands this new obligation:
The strategy must, in particular, specify —
“(a) the funding level the trustees or managers intend the scheme to have achieved as at the relevant date or relevant dates, and
(b) the investments the trustees or managers intend the scheme to hold on the relevant date or relevant dates.”
Of course, this is very far from the straitjacket terms of the proposed TPR Code. The present Bill does not even preclude open scheme trustees from specifying that horizon as, say, thirty years from now, and at times in the future, but…
The Bill also grants powers to the Secretary of State to write further secondary legislation, regulations, and lists some that may be enacted. However, that is for another day, a separate consultation and due Parliamentary process. The Bill does, however, specify:
”(b) “relevant date” means a date determined in accordance with regulations.”
The proposed new DB Funding Code sits at the end of this legislation where it should be remembered that, although it must have been laid before and approved by Parliament, it does not have the force of law. It is guidance, albeit offered in a ‘comply or explain’ framework.
With all this in mind, we avidly listened to, and watched the recent TPR webinar on the proposed Funding Code. We had hoped for some rigorous analysis and justification of the need for a different funding strategy such as self-sufficiency (sorry, low dependency) for closed schemes in run-off. We were disappointed. By contrast, we were pleased to hear that ‘prudence’ will be defined in the second consultation on the Code.
We were surprised by the emphasis on the ’fast-track’ as the reference base (it was even referred to as a benchmark at one point) from which all ‘bespoke’ arrangements are to be viewed. This was described as ”objective”.
Later, TPR’s Head of Legal, while describing objective measurement, quoted the following line from the website of the Institute of Objective Measurement:
”The goal of objective measurement is to produce a reference standard common currency for the exchange of quantitative value, so that all research and practice relevant to a particular variable can be conducted in uniform terms.”
Now, that is a description of purpose rather than a definition.
The Institute’s website does offer the following definition:
“Objective measurement is the repetition of a unit amount that maintains its size, within an allowable range of error, no matter which instrument, intended to measure the variable of interest, is used and no matter who or what relevant person or thing is measured.”
But that, of course, would have forced TPR to face the fact that their principal measure, the gilt-relative discount rate, fails this test. Using a market-based gilt rate, the amount returned by accrual of the contributions and the amount returned by discounting of the projected cash flows are not the same.
Had we known that TPR was thinking in these terms, we would have offered to loan them our three-volume copy of ’Foundations of Measurement[i]’. We would have added the caution that its 1,453 pages are mathematically challenging and that after 15 years of effort, those authors abandoned their work on stochastic measurement, as mathematically too difficult. And, of course, the principal measure in the Code, the gilt-based discount rate, is stochastic.
As this term is often bandied around, and its meaning often unclear, we offer the following two definitions from the Free Dictionary:
- Statistics involving or containing a random variable or process:
- Of, relating to,or characterized by conjecture; conjectural.
These definitions hardly support claims to be “objective measurement”.
The problem is that the value of the gilt yield that we observe[ii] is a probability zero event[iii], and that is hardly the basis for any sound decision. We can and do work with stochastic processes, but when we do so, we use collective statistics, such as the mean, median and higher moments of the process[iv], not arbitrary values from within it. This is the theoretical basis for techniques such as smoothing.
However, for the Regulator there is a far larger problem; this renders baseless any attempt at enforcement based upon these discount rates. ‘Fast Track’ cannot be a valid comparator or ‘benchmark’.
Ordinarily in financial analysis we welcome extreme events as these are information-rich – the further from the mean the richer they are, the more they tell us. However, this information is noise, distorting the values obtained, when used as a measure, as in discounting.
However, all need not be lost. Recognition of this by the Regulator might free us from the tyranny of benchmarks and perhaps even the tyranny of metrics.
Indeed, the elimination of metric-induced herding, goal displacement and short-termism might actually free us to pursue a strategy for providing pensions over the long term.
[i] Foundations of Measurement. Vols. I. II, III. Krantz, Luce, Suppes & Tversky. Dover Publications 1979, 1989, 1990.
[ii] Strictly, we observe a price from which we calculate the yield)
[iii] Readers wishing to understand this point more fully are referred to ”Not all probabilities are created equal” https://www.probabilisticworld.com/not-all-zero-probabilities/ or a good text on measure theory such as:
“Measure Theory and Probability” by Malcolm Ritchie Adams and V. Guillemin,
[iv] We also pay great attention to the error terms of these statistics and the confidence we may place in them.