Universities Superannuation Scheme (USS):

A consultation for the 2020 Valuation[1]

Some General Views

Michael Bromwich [a]

Assumptions are dangerous things to make, and like all dangerous things to make, bombs, for instance, or strawberry shortcake, if you make even the tiniest mistake you can find yourself in terrible trouble.       Daniel Handler

The USS is the pension fund for academics. It is the biggest private sector defined benefit (DB) pension scheme by assets with some 460,000 members offering career average DB benefits up to a salary cap and then a defined contribution element. It has a record of substantial deficits and attempts to reduce them. It is unusually a fully open scheme for a sector which will exist into the foreseeable future.  It illustrates many of the difficulties of DB schemes including those of experienced by long lived schemes.   It is one of the first schemes to adopt self-sufficiency-a low risk/low return   investment portfolio which requires no calls for further contributions as a measure of performance. This is the objective which the Regulator is currently going to require or use as a   benchmark.

This consultation with the Universities UK (UUK), the employers’ organization is formally only on the Technical Provisions (TPs). However, a wide range of illustrative examples incorporating some of the other areas are provided to   give an idea of how the elements of valuation impact upon each other.

This note first considers in a general way the possible TPs presented and then evaluates their affordability. It discusses the assumptions used, the USS’s requirements to obtain a strong covenant and its wish to be able to reach self-sufficiency. Finally, it considers the USS’s monitoring and Action Framework. 

The TPs

The consultation does not give a clear indication of the required TPs rather it gives a medley of possible outcomes. It provides figures for a tending to strong covenant and a strong covenant combined with a variety of investment outcomes with increasing covenant support. Some of the results are shown in the table below.

Summary Outcomes

Table 1: Potential range of Technical Provisions and Associated Contributions

* All figures estimated assuming linearity, **. With payroll growth of 2% and a post retirement discount rate of 1%.  Source: selected figures from the USS’s Consultation for the 2020 valuation, tables 9.1, 9.5 and 9.7.

The table provides an illustration of a general problem with the consultation. Only USS has the personnel data necessary to model university pensions. Lack of knowledge of the detailed logic and of the nuances behind the numbers does not allow detailed investigation of these and other figures. USS given their fiduciary and legal responsibilities often seems unwilling to give further information. It would aid analysis if USS were able to share the architectures of their models.

This problem is accentuated by the use of numerous illustrative models with assumed outcomes for the areas not yet consulted upon to    provide estimates of possible overall outcomes. The weights that will be given to these assumed outcomes in the valuation is not clear even though much of the consultation is devoted to stating for the assumptions used in the models.

The column using gilts+4.5% is added to the USS’s results and is estimated. Aon in its comments on the USS consultation gives its view of the results of using gilts+3.5% with a 15-year recovery plan saying that this would generate a seven percent deficit contribution and a total contribution of 36.4 percent.[2] Allowing a 15-20-year recovery plan as suggested by the Joint Expert Panel would similarly reduce further the total contribution associated with the gilts+4.5% investment strategy shown in the table[3].

The USS also gives results allowing for a 0.5 percent investment outperformance in the recovery plan and for a ten-year recovery plan. This generates 12 different total contributions ranging from 67.9 percent to 45.4 percent. These are ‘zombie’ contributions as they all are impossible for universities to fund without of major damage to their finances and activities. What management theory supports at least 45 percent of salaries going into pensions and what   types of employees would desire this (even with tax allowances)?

It is difficult to avoid the view that the current approach is:

‘Here are the results, what is the Joint Negotiating Committee which determinates benefits going to do about them’.

Some commentators have suggested that these figures are aimed at ‘nudging’ universities into accepting the measures the USS requires to ensure a strong covenant.

The table illustrates the high sensitivity of the outcomes, deficits and future service costs, to changes in assumptions about the covenant. These outcomes are also sensitive to the assumptions made in determining their values, such as the length of the recovery plan.  These values also feed into other calculations which also have their own assumptions.  All valuation assumptions are a matter for USS subject to consultation with UUK. The only ways to mitigate the effect of these assumptions when confirmed is for the Joint Negotiating Committee to alter benefits and possible industrial action by UCU.

Affordability

The regulator’s current funding code for DB pensions makes it clear that affordability needs to be considered by trustees and this is redoubled in the new code which is subject to current consultation.[4] During the negotiations over the 2017/2018 valuations both the employers’ organisation UUK and the   College and University Union (UCU) set red lines on affordability, 18 percent of pay for UUK and for UCU eight percent coupled with a no detriment on benefits. Currently they are paying a yearly total of 30.7 percent (employees 9.6 percent and employers 21.1 percent) rising to 34.1 percent (employees 11.0 and employers 23.1) in October 2021.The current cost to employers is regarded by them as the maximum that can be afforded, reiterated in the UUK ‘s response to the USS consultation[5]. It is understood that the union adheres to its red lines.

Based on their view of employers’ risk tolerance and affordability the USS requires that the employers can be called upon to provide if necessary up to 10 percent of payroll per year rising in line with the CPI +2 percent per year for 20 years for affordability issues with deficit contributions in sustained adverse scenarios with a tending towards strong covenant and 30 years with a strong   one. This is labelled affordable risk capacity. In the UUK response 44 percent of respondents said all the outcomes presented by the USS   were either   not affordable or    above affordable risk.  An additional maximum call of 15 percent per year with the CPI adjustment for 20 or 30 years depending on covenant strength is required to protect from   extreme issues not allowed for in the TPs. The total of the two risk calls together are labelled available risk capacity. In the UUK response 18 percent of universities agreed to a 25 percent call most only for a short period    whereas 39 percent said they could not sustain further  increases in contributions.

The USS is clear that that a sustained 10 percent call could involve employers changing their plans and may involve some employer failures.  They do not give examples of   the settings they have in mind as   causing affordable risk calls.  Calling available risk capacity could involve   failures over a significant proportion of the sector and the   possibility of further harm to the sector accompanied by moving to self-sufficiency. It also imposes extra pressure on employers already dealing with extreme issues. No examples of such situations are given except where the scheme must close to future accruals.  This lack of   indications of the extreme settings in mind and their probabilities deprives employers of any ability to evaluate the possibilities of calls and their costs and to mitigate them.   A similar stance was taken by USS in the 2018 valuation regarding to contingent support. It is not clear that the figures in Table 1 are due to such anticipated adverse scenarios. They rather seem to flow from the assumptions.

The required contributions in the table are way beyond the avoidable risk capacity and would either destroy the sector’s ability to be able to answer these calls for additional funds or require substantial changes in its plans in situations which USS recognises will already be disrupting universities. It does not seem rational to present such figures without mitigating suggestions.

This approach privileges the scheme above the sector.  As written, it flies in the face of the Regulator’s exhortations for cooperation between sponsors and trustees and ignores the regulatory requirement to sustain growth.

For the first time USS has indicated that its maximum risk appetite is for risk that when crystallised would require a maximum call of 10 percent of payroll per year upon employers. Currently this would be over-exhausted by the contributions shown in Table 1.

Assumptions

Underlying the figures are some major changes to the 2018 valuation assumptions. The USS’s   results include two of the Joint Experts Panel’s risk-increasing recommendations. They incorporate dual discount rates and on the face of it have abolished Test 1 which generated automatic de-risking. They rejected the recommendations to smooth future service costs over two valuations and to allow long recovery plans as both are regarded as generating additional risk and to use the consumer price index rather than gilts.   Further consideration of the above assumptions may be helpful.

This illustrates a fundamental problem when seeking to value DB pensions. These valuations require estimates of the future including the distant future. Valuations are thus constructs, that is they are not based on empirical evidence. Necessary empirical evidence is generally unavailable including the market prices of long-term liabilities. Markets for these are generally thin and imperfect. Thus assumptions must be made and will differ between individuals. Valuations will involve battles about assumptions. Only empirical experience can say eventually who if anyone was correct. Favouring one valuation (construct) over another is a choice between assumptions or beliefs not facts.

Under the scheme’s rules the USS holds has all the cards in   setting assumptions as these are a matter for the   Trustee on the advice of the scheme’s actuary subject to consultation with UUK.   Chosen assumptions can be enforced   by the USS through cost sharing if the Joint Negotiating Committee cannot reach agreement on benefits or restructuring the scheme.

Although the UUK provided detailed comments based on universities’ responses to the USS’s consultation there have yet been no detailed rejoinders to the UUK.   Any changes will be in a paper in early 2021 to the Joint Negotiating   Committee setting out the scheme’s financial condition and indicating the required   employers’ contributions to sustain the existing benefits. The Regulator has provided views on these matters. Additional scenarios   will also be provided to meet UUK’S request for information sharing[6]. There would seem to be a need to have better system to generate shared assumptions.

 Detailed Assumptions

In addition to the accepted Joint Expert Panel’s assumptions there are other crucial new assumptions.

 Prudence

USS indicates that more risk has been allowed in the 2020 valuation than in that of 2018. A weighted or single equivalent discount rate gives an overall rate for the pre-and post-retirement discount rates. The single equivalent discount rate for the USS’s highest investment rate in Table 1 is gilts+1.9 percent whereas those for the 2018 and 2017 valuations are gilts+1.33 percent and gilts+1.2 percent. The latter two rates are poor comparators for determining  the degree of   greater risk taking as both incorporate major de-risking in their first ten years.

However, the deficits and future service costs in all cases are higher than those at the last valuation on 31 March 2018 (deficit of £3.6bn and future service cost of 28.7percent).  USS says these outcomes are driven by the deteriorating outlook for future investment returns and the need for greater prudence. USS warns that “Taking more investment risk would therefore probably be accompanied by a greater need for prudence in the discount rate. As a result, if we were to assume a higher risk investment strategy than proposed, this would not necessarily lead to lower contributions”. USS says that such additional risks require more prudence. This should be the case only if the probability distributions of the more expensive investments have changed their characteristics.

Prudence in the 2020 valuation is increased as the confidence levels applied to returns are higher than in 2018 by over ten percent (therefore the discount rates are lower). A higher confidence level reduces the levels of investment returns that can be considered and  be actioned  in the discount rate. The confidence level applied in the 2018 valuation was 67 percent but in 2020 with a strong covenant the required confidence level for pre-retirement returns is 78 percent and the post-retirement level is 73 percent. The former level rules out a return of gilts+4.5 percent whereas this return was available in the 2018 valuation.  The USS’s reasoning for these higher levels lacks any precision. It is just a value judgement as it currently stands.

An alternative approach

An alternative approach to prudence is to determine the optimal proportion of growth assets given the risk attitude employed and then increase the portion invested in low return-low risk funds to reflect the desired prudence. This approach would be more transparent than the conventional approach.

The 2020 valuation incorporates a substantial amount of extra prudence including a short recovery plan, no allowance for outperformance, required additional support for a strong covenant, lower returns for deferred members with a weak covenant and much higher reliance on employers. That the USS can impose cost sharing further heightens prudence as does achieved de-risking. This portfolio of additional prudence-generating instruments suggests extreme conservatism rendering unnecessary the higher confidence levels in the 2020 valuation.

Aon estimates that the illustrative investment strategy incorporates initial de-risking will cost seven billion pounds; additionally the abandonment of the reversion of gilts prices costs four billion pounds (USS estimate) and moving if necessary to self-sufficiency some 35 billion pounds. (six billion per year (USS estimate).

The Covenant

USS puts great emphasis on the covenant and models a tending to strong covenant saying that a strong covenant would require very specific additional support from employers. The Regulator although wanting covenants to be strong enough to support plans prefers a covenant-free approach in determining the amount of growth assets held depending on a scheme’s maturity. More mature schemes are seen as being less flexible when dealing with   volatile and weak investment outcomes. This illustrates the importance of different assumptions.

Growth Assets

USS   illustrative investment policy with a strong covenant is for active and deferred members to hold 90 percent growth assets and for pensioners 90 percent of low return-low risk assets. This policy gives an average holding of growth assets of 55 percent relative to 65 percent in the 2018 valuation. This stems in part from the view that all pensioner payments from the date of members’ retirement are to be funded from low return assets at retirement.  This seems overly prudent.

With a tending to strong covenant the average holding of growth assets is 40 percent reflecting a lower tolerance to risk stemming from a weaker covenant. This change relative to a strong covenant is also generated by assuming that non-retired deferred members funds are invested in low-risk investments. No reasons are given for this approach. Both average figures reflect de-risking that by year 20 will reduce their holdings of growth assets by five percent.

 Length of Covenant

USS following its advisors takes the visibility of a strong covenant as 30 years and of a tending to strong one as 20 years. This difference in visibility is difficult to understand. The university sector will not disappear after 20 years. Its shape may change–there may be some amalgamations. The financial strength of the sector may be weaker with a tending to strong covenant but this should be factored into expected returns rather than reducing visibility.

With the USS’s   approach the strength of the covenant is crucial.  It determines together with the employers’ and the USS’s risk attitudes-the reliance that USS can place upon employers over time for additional contributions. The stronger the covenant the more flexibility is available to deal with risk.

They accept that a strong covenant allows the length of the recovery period to be extended. The Joint Expert Panel recommended 15-20 years for the USS scheme. Currently the USS is using 8 years (the remaining length of the 2018 recovery plan) as required by the Regulator. Allowing long recovery periods is in complete conflict with the Regulator’s views. The Regulator believes that the visibility of covenants is generally limited to three to five years and the length of recovery plans should be consistent with this subject to a six-year maximum. Moreover the Regulator takes the view that schemes with strong covenants should be able to settle recovery plans quickly except for open schemes that are willing to provide legally backed ‘insurance’. Applied to USS   this would place substantial extra costs on employers.

This seems to misunderstand the general business model of universities. Universities annual surpluses can be characterised at a high level as based on a cost-plus model with a relatively small margin over cost even with a strong covenant. The annual free cash flows of universities are   constrained by the levels of revenues (fees and research funds), operating costs and planned investments. It is therefore difficult for universities to meet large up-front lumpy payments whereas payments spread out over a long period are more likely to be sustainable.

Strengthening the Covenant

In the 2018 valuation the USS made suggestions for strengthen the covenant from tending to strong which are still being negotiated and have been strongly urged in the 2020 valuation. These suggestions are to allow USS to monitor the university sector’s debts, give pari passu status with new university debts and to require employers who wish to leave the scheme to obtain permission from the USS.  A review of the covenant in November 2020 by the USS’s advisors said that the sector could support a strong covenant with the right support package.

Illustrations in the consultation suggest that the consequences of having only a tending to strong covenant are severe including that the expected returns are only one percent above those of a low-risk-low-return portfolio (a self-sufficient portfolio), a covenant visibility of only 20 years, deferred members are assumed to be invested in a self-sufficient portfolio and the proportion of growth assets held is   limited to 40 percent reducing to 35 percent over time.  The logic behind these effects is not presented (see above) but as indicated clearly increase prudence.

The USS suggest that a strong covenant would put the return  in the middle of the presented discount rates leaving the total contribution still unaffordable. This seems surprising.

The covenant support requirements represent a substantial and an everlasting reduction in universities’ sovereignty. Given the amount of prudence which seems to be employed in the consultation these restrictions would seem unnecessary and are without any guarantee that they would lead to a sustained strong covenant.  Without ring fencing they are likely just to be baked into future contributions as happened with contingent contributions in the 2018 valuation.

Self-sufficiency and Monitoring

In past valuations self-sufficiency has driven the need for de-risking via the notorious Test 1. Following the suggestion of the Joint Expert Panel this test has been abandoned for the 2020 valuation but self-sufficiency has not gone away.  It provides what USS  see as a safe harbour in extreme circumstances and they   wish to maintain access to this within a 10 percent call on employers.  When achieved self-sufficiency provides near certainty of payments of benefits as they fall due without further support from employers.  Such an achievement comes at high cost both in terms of buying currently expensive gilts and in foregone   returns from growth assets. USS  indicates that the current cost of the required de-risking to attain self-sufficiency is some 35 billion pounds.

In accordance with the recommendation of Joint Expert Panel the distance between the TPs and self-sufficiency is now seen only as a risk measure but it is far more than this. For example, a call on employers’ resources could be expected if both current and future self-sufficiency deficits are near or above available resources and could be accompanied by a move to self-sufficiency.

This continuing emphasis is made clear by target for the gap between TPs and self-sufficiency over the next 20 to 30 years being set depending on covenant strength between   22 billion pounds and 38 billion pounds.  This target is not justified by USS. The reliance that USS is willing to place on employers is   much greater than in previous valuations but this does not seem to be reflected in the illustrative contributions. When TPs deficits are published these are usually accompanied by their self-sufficiency equivalents so that self-sufficiency can be seen to be being kept within reach.  It is difficult to avoid viewing self-sufficiency as at least a ‘shadow’ long term objective. This seems to be consistent with the Regulator’s treatment of open schemes. Such schemes can avoid having a self-sufficiency long term objective providing that they are able to furnish very strong evidence of sufficient resources to avoid consequent problems.

This   view of self-sufficiency is strengthened by considering the   monitoring and action framework presented in the USS’s consultation which is strongly self-sufficiency orientated.  it is not being consulted upon at the present time being presented only to help understand the presented TPs.

Such monitoring frameworks are important as they are both published and shared with Regulator. The monitoring framework generates variances between the self-sufficiency target and available resources. The 2020 framework is focused entirely on the ability to access self-sufficiency. This framework is substantially different in kind to that   associated with the 2018 valuation which was more operational[7].

The monitoring framework has three metrics each with three thresholds.  The first metric   is long-term orientated and compares affordable risk capacity (AR) with the amount of the self-sufficiency (SS) liability not covered by the TPs which are assumed to be fully funded. The degree to which the available amount of AR covers this difference is called the head room.  The thresholds are the minimum degrees of coverage required under each metric. There are three definitions of head room for each metric ranked as green, amber and red depending on the areas of risk that must be covered mainly that of the risk attached to the transition to self-sufficiency and extra demographic risk. Metric 1 is highly reminiscent of Test 1.  The only obvious way of remedying an adverse variance (red) in the headroom is by increasing the TPs by de-risking with requirement of consequent extra funding. Amber indicates that the two major risks are marginally covered and green that a margin exists.

The second metric is the USS’s measure of short run risk. It asks whether AR is sufficient to bridge the difference between the SS liability and the value of assets held at valuation.  Here remedying a red signal would mean higher contributions and possibly a shorter recovery plan.  It is difficult to see how this measure is short term.

The third metric is concerned with extreme downsides and the emphasis is therefore on the ability to move to SS. The metric asks whether the full risk capacity of 25 percent of payroll is sufficient to cover the difference between SS liability and asset value. Red would require   a view on whether moving to SS and de-risking would be necessary as would amber.  Here the head room   is required also to cover the cost of moving to SS and a ‘value at risk’ calculation to capture return volatility[8].  This metric is green for all three thresholds.

The flags of different colours  suggest a dashboard but most dashboards provide indications of the current conditions of a number of parameters. Red usually means something needs to be done immediately to remedy a problem with one item. Amber provides   a warning of the likelihood   of a current problem, again for one item.   The USS’s risk metrics are focused on only one item the ability to achieve self-sufficiency and therefore it perforce takes a long-term view but problems may arise in any time period in the process not just at the time of valuation.

The monitoring system rather than being a dashboard may be better thought of as like a long-term weather forecast. Both models   are mainly affected by exogenous factors beyond the control of the actors. Long term weather forecasts have generally been abandoned   due to inaccuracy.  Mitigations are the only actions available for likely downturns. The outcomes of the predicted settings and the results of any mitigation are uncertain.  USS   have not yet said what actions they will take in the light of variances though they have said that they will not go to self-sufficiency unless several institutions collapse.

Modern management control theory   suggests that variances indicate either the need for better prediction or better performance. The variances due to exogenous factors   are seen as learning opportunities indicating whether predictions need to be reconsidered. USS does not publish anything about the accuracy of its predictions. Rather it rolls these into its explanations of how deficits change between valuations.  It does however require funds up front to meet the effects of changed predictions. It surely should indicate how accurate its predictions are relative to actual experience at least for past years.  Management control theory considers how well the organisation has done ex post each year relative to yearly targets (budgeting) thereby indicating progress towards organisational objectives.   With the current process as exemplified by the USS’s monitoring and action framework there is no reference to    empirical evidence except for asset values.  Even with investment performance which is reported upon separately performance is not benchmarked against the usual market performance indices but rather against the   USS reference portfolio.

Management control theory would also suggest that the instruments in the monitoring framework are highly aggregated and therefore a great deal of   information is lost relative to a more granulated approach.   Deficits are the results of the interactions of many variables which are hidden by the monitoring framework. Management    control theory would suggest that managers (sponsors) should be able to drill down to these variables. This does require the use of   the short-term budgeting system used almost universally in organisations. No doubt USS uses such a system internally.

Many of the above problems arise from the lack of an agreement as to the long run objective(s) of long term or fully opened schemes. A number have been suggested but not really supported in the actuarial industry. The white paper prior to the 2019-21 Pension Schemes Bill said the objective was to ‘run on’.  The Regulator and USS have self-sufficiency in mind. Others prefer a cashflow based approach and others the rate of return necessary to pay what is promised[9].   A suggestion based on this note would focus on the fundamentals of all schemes the cashflows and determine the TPs with a 20-or 30-years’ decision horizon with a discount rate based on the long-term portfolio it is planned to hold in steady state with the journey planned to this point monitored over time. The reliance expected to be placed on employers could be calculated on this basis as could the resilience of the scheme to shocks via scenario testing. Alternatively, it could be argued that the discount rate should be based on an estimate of the marginal cost of capital of employers.

[1] Universities Superannuation Scheme, a consultation for the 2020 Valuation: A consultation with Universities UK on the proposed methodology and assumptions for the Scheme’s Technical Provisions, Universities Superannuation Scheme, 7 September 2020.

a CIMA Professor of Accounting and Financial Management Emeritus, London School of Economics and Political Science.

[2] Aon, Universities Superannuation Scheme, prepared for Universities UK, Aon, 24 September 2020.

[3] A panel set up by the employers’ organisation  and the union to make recommendations on methodology and governance following the 2017 valuation.

[4] Defined benefit funding code of practice consultation, The Pension Regulator, 3 March 2020.

[5] USS Employers, 2020, The USS Trustee’s Consultation on Technical Provisions for the 31 March 2020 Actuarial Valuation USS Employers Response, USS Employers, 13 November.

[6] USS  2020 valuation update, Universities Superannuation Scheme 18, December 2020.

[7] The monitoring and action framework   associated with the 2018 valuation was more focused on the behaviour of operational aspects of the valuation featuring a self-sufficiency deficit affordability ratio, a future service cost coverage ratio, a deficit recovery contribution adequacy and covenant ratings.

[8]  Appendix E of the USS consultation gives a quantitative illustration of these metrics with a strong covenant and a discount rate of gilts+3.5%.  Metric A is marginal between amber and green. Metric B is marginal between amber and red and metric C is green. These gradings are very sensitive to changes in assumptions and the thresholds   It is difficult to understand the derivation of the thresholds.

[9]  See Clacher, Iain, Keating, Con, & Tilba, Anna. 2017. A Primer on The Risk Structure & Contractual Accrual Rate of DB Pensions, Long Finance, October.

Michael Bromwich