On USS’s discount rate and long-term funding targets
Number 77: #USSbriefs77
This is a USSbrief, published on 19 August 2019, that belongs to the OpenUPP (Open USS Pension Panel) series. It has been submitted by the author to the UCU-UUK JEP (Joint Expert Panel), with the title ‘Comments on scheme valuation and associated matters’, on 30 June 2019. The author is a USS scheme member, a member of the University of Sheffield USS Working Group, a UCU-elected JNC member (June 2018-present), and is writing in a personal capacity.
1. The importance of stability in the discount rate
Questions surrounding the sustainability of a Defined Benefit (DB) scheme essentially come down to the affordability of the contribution rate (including deficit recovery contributions) determined at each full valuation of the scheme. By far the biggest determining factor is the assumption regarding the discount rate. It is instability of the discount rate from valuation to valuation that is the primary driver of instability in the contribution rate.
I do not have a firm proposal for how to achieve a stable discount rate, though I do not believe that aiming for such a thing is impossible or at odds with the legislation, which allows the discount rate to be chosen taking into account ‘the yield on assets held by the scheme to fund future benefits and the anticipated future investment returns’ (Pensions Regulator, 2014). If the asset mix held by the scheme changes relatively slowly, then, without a step-change in the prevailing long-term economic model, it would seem possible for the discount rate to adjust slowly to take into account recently observed market conditions.
One could, for example, start from an evidence-based position of long-run, observed returns on the asset classes, adjusted for prudence and possibly subject to a pre-agreed cap, which could then be adjusted up or down gradually as more data is collected. Where there is clear evidence that this approach is likely to give an overly optimistic view of future returns (for example, where there has recently been a major shift in economic policy), the margin for prudence could be increased and/or a cap on the rate could be applied.
Triennial valuations coupled with the concept of deficit recovery means that pre-empting a fundamental shift in future returns is not all that materially different from observing the same shift after it has begun to occur: the difference will be at most three years’ worth of underpayment.
2. Long-term funding targets
2.1. The approach taken by Test 1
I have written extensively on USS’s Test 1 (see USSbrief32, USSbrief51, USSbrief58, for example), which sets a target for the Year 20 technical provisions liabilities at a fixed distance from the Year 20 self-sufficiency liabilities, based on a pre-determined ‘reliance on covenant’. The effect is to dictate a change in investment strategy towards low-return assets. I will not repeat earlier comments, other than to say that:
- I do not believe that the Test 1 algorithm is necessary (or, for that matter, a good way) to ensure long-term sustainability of the scheme.
- I consider the current application of the underlying ideas to be causing damage to the scheme’s longevity and unhelpful to it achieving full funding.
I attach a document written to aid discussions between the stakeholders of the scheme in Appendix A, where the main conclusions drawn from prudently adjusted Year 20 projections provided by USS are outlined. In particular, by targeting a figure for the technical provisions rather than assets at Year 20, the algorithm makes both full funding on a technical provisions basis and the targeted reliance on covenant at Year 20 harder to achieve.
I do not believe there is any merit in retaining Test 1 in its current form in the valuation.
2.2. An alternative approach
The Joint Negotiating Committee (JNC) was recently provided with modelling of projected Year 20 self-sufficiency deficit distributions under three scenarios:
- Where no de-risking occurs
- Where de-risking as mandated by Test 1 with a £13bn reliance is carried out in Years 11–20
- Where the same Test 1 de-risking is carried out in Years 1–20.
This modelling was based on a total contribution rate of 20% (employer plus employee), and comprised 2,000 trials in each scenario.
While the number of trials was insufficient to demonstrate any clear positive effect from de-risking over the 20-year timescale (see (5)), the format of the modelling would make possible an alternative approach to dealing with the same underlying long-term funding target, namely to determine the contribution rate at which the probability of the Year 20 self-sufficiency deficit exceeding an agreed ‘reliance on covenant’ is sufficiently low. 
Such an algorithm would not drive investment strategy but could be used to inform it: by seeing how the contribution rate determined by the algorithm changed as the investment strategy changed, one could make informed decisions on the preferred approach. It would also not drive the valuation, but instead give a separate indicator as to the level of contributions needed to achieve the specified long-term funding target with a sufficiently high probability.
Marsh, Sam (2018) Understanding ‘Test 1’: a submission to the USS Joint Expert Panel. USSbriefs32.
Marsh, Sam (2018) Addendum to ‘Understanding “Test 1”: a submission to the USS Joint Expert Panel’. USSbriefs51.
Marsh, Sam (2018) A flawed valuation: the layperson’s guide to my findings on USS’s ‘Test 1’. USSbriefs58.
Pensions Regulator, The (2014) Code of Practice 03: Funding Defined Benefits.
Appendix A: Test 1, and the implications of the Year 20 asset projections
A discussion document prepared by UCU’s Superannuation Working Group, October 2018 
There is much criticism of Test 1 in the first report of the Joint Expert Panel (JEP). This discussion document explores those criticisms more fully in light of the recent Year 20 asset projections sent by USS to JNC member Sam Marsh. These projections give the most complete picture we have so far of how Test 1 interacts with self-sufficiency and the reliance on covenant, and illustrates that Test 1 may not be fit for its intended purpose, with more information urgently needed to make an informed view.
The evidence now available demonstrates that the current commitment by UUK members of 7% of payroll over 20 years is more than enough security to bridge the gap between the prudently projected value of the assets and self-sufficiency at Year 20 before any de-risking occurs. This allows self-sufficiency to be considered in a more appropriate fashion, in line with comments from the Joint Expert Panel. The conclusion is that continuing with current benefits and contribution rates may carry less risk than previously thought, and has the potential to form a mutually beneficial outcome for the 2017 valuation.
JEP criticisms of Test 1
The following is a non-exhaustive summary of the Panel’s criticisms of Test 1.
- Test 1 is ‘highly sensitive to the input assumptions’ and ‘being used as a constraint on benefit design and driver of investment strategy’; the Panel ‘does not consider this helpful’ (p. 8).
- ‘The way in which the employers’ risk appetite has been applied through Test 1 has contributed to the adoption of strong risk aversion’ (p. 8).
- ‘Test 1 has led to a valuation that is model-driven rather than model-informed’ and ‘alternative ways of arriving at a valuation of technical provisions are open to USS and should be explored’ (p. 24).
- ‘The Panel agrees with some other commentators that there are other ways of calculating the value of reliance, for example by monitoring the difference between the self-sufficiency liabilities and the Scheme assets’ (p. 31).
- ‘USS’s approach to meeting Test 1 implies a de-risking of assets. A number of other paths are open to USS and could be explored’ (p. 31).
Note also the comments on p. 29, where the Panel’s views on self-sufficiency are that:
- ‘Self-sufficiency is a useful concept. It provides a reference point for judging the extent of reliance on the employer covenant and whether the Scheme has sufficient funds either from its assets or that can be called upon from employers at any point in time’.
- ‘However, the way in which self-sufficiency is used drives up the deficit and contribution rates. The Panel believes that very few schemes use self-sufficiency as an input to the valuation of technical provisions in such a formulaic way’.
Year 20 projections
The Year 20 projections provided by USS give an indication of the extent to which the de-risking limits the future performance of assets and the expected Year 20 ‘reliance on covenant’ under three different investment approaches.
The figures below are based on the prudent discount rate used by USS in their valuations (not best-estimate) and assume the scheme stays open with unchanged contributions and benefits. All values are adjusted for CPI inflation.
* Sam Marsh’s calculation based on USS cashflow data and confirmed to be ‘reasonably consistent’ with USS’s own calculations.
† Determined as £81bn–£10bn, as required by Test 1.
Other data as stated by USS and adjusted for CPI where appropriate.
- Before the application of Test 1 and the de-risking it forces, the scheme is expected, on prudent assumptions, to find a Year 20 surplus of around £19bn in real terms and a ‘self-sufficiency deficit’ (or ‘reliance on covenant’) of only £2.8bn.
- Not only is there a rise in the technical provisions liabilities for both de-risking approaches, but also a fall in the expected level of the assets.
- The scheme has an expected Year 20 surplus under the September assumptions: the current contribution rate is more than enough for the scheme to be fully funded by Year 20.
- Under the November assumptions, the current contribution rate is expected to result in a small deficit. 
- From this table we can derive that the probability of a reliance on covenant of £10bn or greater developing is a) less than 33% with no de-risking; b) around 33% with the September de-risking; c) greater than 33% according to the November de-risking. 
- We have no information of how the probabilities of finding a Year 20 ‘self-sufficiency deficit’ of any larger size (£20bn, £40bn, etc.) vary according to the different de-risking plans. This would be key to making an informed decision on risk appetite.
- It is known that, with no de-risking, the deficit as at 31 March 2017 would be around £0.5bn, with future service costs likely to be around the current contribution rate.
We ask UUK to join with UCU in expressing deep distrust of Test 1, and explore whether it is possible to reduce its impact on the 2017 valuation.
 It is hard to know, without seeing appropriate modelling, what one should deem as a ‘sufficiently low’ probability. From the modelling provided, it would appear that a total contribution rate of 20% with no de-risking would give around a 30% probability of the self-sufficiency deficit exceeding £13bn.
 The figures in this document have been lightly amended from those in the original to reflect the fact that the numbers confirmed by USS are not in the public domain. These numbers are ‘reasonably consistent’ with the ones presented here, and the conclusions are unaffected.
 Note: the figures in the table give this deficit as zero, though the numbers confirmed by USS give rise to a small (less than £3bn) deficit.
 This conclusion follows from the numbers supplied by USS.
This is a USSbrief, published on 19 August 2019, that belongs to the OpenUPP (Open USS Pension Panel) series. It has been submitted by the author to the UCU-UUK JEP (Joint Expert Panel), with the title ‘Comments on scheme valuation and associated matters’, on 30 June 2019. The author is a USS scheme member, a member of the University of Sheffield USS Working Group, a UCU-elected JNC member (June 2018-present), and is writing in a personal capacity. This paper represents the views of the author only. The author believes all information to be reliable and accurate; if any errors are found please contact us so that we can correct them. We welcome discussion of the points raised and suggest that discussants use Twitter with the hashtags #USSbriefs77 and #OpenUPP2018; the authors will try to respond as appropriate. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.