If you couldn’t be at the CISI 24 Pensions forum, you missed this great speech from Jackie, thanks to the wonders of google, this blog can bring you not just her graphs, but a transcript of what she said on the day
Hi Everyone, My name is Jackie Grant. I’m a Senior Teaching Fellow in physics at the University of Sussex and also an elected national negotiator for UCU, the trade union representing around 120,000 staff at UK universities. This means for nearly two years I’ve been a member of the Universities Superannuation Scheme’s Joint Negotiating Committee that among other things sets benefits and contribution rates for our pensions.
It’s a real pleasure to be here and be among such esteemed company – I have to admit it’s not somewhere I ever thought I would be, because I only started to seriously think about my pension in 2017 when I was faced with a decision on taking part in very prolonged industrial action and so having to consider the impact this would have on students and how I would respond to their questions.
So I plan to briefly talk through a somewhat personal view of the history of USS and the most recent years of industrial disputes.
As I’m sure many of you know, the UK Higher Education sector has been in dispute over USS pensions for over a decade and in particular the last 6 years has seen this dispute become increasingly acrimonious, with widespread industrial action of many forms including 69 days of strikes. All to the detriment of the hundreds of thousands of staff, millions of students and, I believe, the international standing of the entire UK University sector.
So this first graph I want to consider shows three sets of USS assumptions that underpin the calculations of pension costs as they have evolved over this period of industrial dispute from 2011 to 2021.
- The black line shows asset growth from 20 to 80 billion.
- The other lines that change in colour through the years show
- firstly: USS projections for their best-estimate of asset growth – these are shown as dots
- second USS break-even growth ie the growth that would be needed from the assets to allow pensions to be paid in full – shown as dot dash
- and thirdly USS choice of ‘prudent assumptions’ – shown as dashed
It is noticeable that the assumptions are getting more and more pessimistic every year.
Importantly it is the lowest line of each year – the prudent assumptions – which is the discount rate that sets the contribution rate and means that since 2011 we’ve seen the associated ‘projected costs’ move from 22% to an eye-watering 49% of salary – and ‘modelled deficits’ move from £3bn to £15bn. With every valuation in this period accompanied by cuts in benefits.
For further context not shown in the graph: the current costs as at March 2023 are now very significantly less pessimistic. Costs are around 20% representing the lowest employee rate of 6.1% since before 1975 and a surplus for the first time in over a decade reporting in at a significant £7bn.
So we are now in the unfortunate situation that Universities are collectively paying over half a billion a year for a ‘modelled deficit of £15bn when the scheme is actually recording an impressive surplus of £7bn.
Myself, my colleagues and students widely consider this a massive waste of resources, not just financial but of staff and student time and energy. Representing an unstable, highly visible and enormous opportunity cost to this critically important sector.
And when it became clear in 2021 that there was going to be no reconsideration of these excessively pessimistic assumptions, I spent a lot of time lobbying my own employer.
The body representing employers is Universities UK and I think I have read everything that UUK has published on USS: and at the time there was a lot that myself and others felt was missing in terms of qualifying assumptions.
In particular a repeated claim from UUK was that their proposal for cuts to pensions to address this modelled deficit was a headline reduction of only 12% for staff on salaries of £40,000 or less.
These figures and others were being sent to University Governing bodies to argue in support of significant cuts to pensions. At our university they were defended by the VC at the time as being a robust and accurate representation of the cuts. So when the statutory USS consultation modeller was made available I worked with colleagues to run the source code over the distribution of staff by age and salary.
The results showed a starkly different picture. In particular for 84,000 staff on salaries below £40k, as shown here – the cuts are clearly significantly larger than 12% claimed by employers. The cuts are larger still when all salaries are included. In fact they show an average of around 35% cut to future defined benefits.
So something critically important was missing in the analysis and communication of the data being used by employers to inform decisions. The subsequent erosion of trust was highly instrumental in yet again pushing us into dispute.
Returning to USS, as the cuts were being imposed in 2022 the USS annual report showed only around 17%, so less than one in five, of USS members as having a good relationship with the scheme.
A repeated theme in discussions I was hearing was the low levels of trust in USS’s own data and analysis.
In order to try to understand what was happening within the USS modelling that was driving me and colleagues to the picket lines on such a regular basis. I made use of the USS quarterly ‘monitoring data’ – published on their website to show ‘direction of travel’. I plotted every single data set against every other data set. The strongest correlation was between the USS modelled cost of future pensions and their chosen value of long-dated gilt yield.
You can see from this figure that it’s a very high correlation – basically the gilt yield value predicts costs to within 1 percentage point; it’s similarly high for other key parameters: the Technical Provisions liabilities and the self-sufficiency deficit.
I’m going to take the opportunity to quote from Erica Thompson’s paper on ways to Escape from Model-land
The first, and most effective, [way to escape from Model-land] is by repeatedly challenging your model to make out-of-sample predictions and seeing how well it performs.
I wonder if the USS monitoring data was ever even tested by running over this range of gilt yields to see if indeed it was the only parameter needed to estimate the modelled costs.
I think if USS continue to use such modelling we may be in a position to predict the trajectory of UK university industrial disputes directly from the gilt-yield.
The final graph here I will only pass over briefly. But this is the area I am currently most interested in because I think this may inform industrial relations over the next three years.
This relates to the USS calculation of the self-sufficiency liabilities. This is defined in various ways in the USS materials but most recently as: a low-risk investment strategy for funding the scheme if the employers are no longer able to support the scheme – or in the absence of a covenant.
These self-sufficiency liabilities are extremely important for USS modelling – because for the 2020 valuation these are the quantities that set the majority of the deficit and were critical for setting the costs.
In addition the Pensions Regulator considers the self-sufficiency deficit important in considering covenant characteristic or strength of employer support and also important in considering the appropriate assumptions and flexibilities for the discount rate.
So it really matters what value the self-sufficiency liabilities take. So I really want to know how these self-sufficiency liabilities are calculated, what assumptions are being used – whether these assumptions are appropriate and if (as I suspect) they are driving the almost perfect gilt-yield dependence.
This graph is the only publicly available data that I know of, showing path modelling of the USS second self-sufficiency condition. There is a lot more I want to say about this and particularly to explore the way it interacts with the presentation we heard earlier today about escaping from model land.
But returning to the long years of industrial dispute.
There is good news: USS pensions will be restored to the level they were before the 2022 cuts and this will almost certainly happen on 1 April 2024. In addition the reduced benefits of the two years since the cuts will be augmented at a cost of around 1 billion.
The contribution rates will be dropping significantly and potentially as early as January – saving staff around 3.7 percentage points of their salaries. This is significant given the 25% erosion in university pay over the last decade and the ongoing cost of living crisis. In addition, it will save the University sector around a billion pounds a year – these are very much needed resources for this strategically important sector.
However, I am still worried. I am worried that if the modelling used so far continues as it has been used, then we will almost certainly be back in dispute if or when gilt yield drops.
I do also believe that there is hope that changes are possible and even already in process. As articulated in the Joint statement from employers and employees
We are committed to work together to ensure the stability of benefits and contributions at future valuations.
And we will work to
… achieve a moderately prudent evidence-based valuation as a driver of stability in time for the next valuation
I am also hopeful that within USS there is both an opportunity and I really hope a willingness to open up and interrogate much of the analysis – so perhaps we could be seeing the beginning of the end of the cycles of disputes. I really hope so – because I don’t want to be back on the picket lines.
Thank you. I’ll end there.