Uncertainty and the USS “deficit”: the curse of false precision

Central to arguments for reform of USS benefits have been claims that the scheme is in substantial deficit. Most of the public argument has been over the size of the deficit, with the £5.1 billion figure from the September 2017 valuation demanding less drastic and damaging changes than the £7.5 billion figure produced by accelerated de-risking in November. Less has been said about the fact that there are many different kinds of deficit, and about just what kind of deficit is under discussion. But this is equally crucial, since the way in which the USS “deficit” has been assessed in fact offers a practically irrelevant and therefore unhelpful way of assessing the financial health of a scheme like USS,  backed as it is by the covenant of around 350 stable and often very large employers.

For most, the word “deficit” first brings to mind the government budget deficit which was central to political debate between 2008 and 2016. This refers to the extent to which public spending in a given year exceeds government revenue. It is therefore important to note that the USS “deficit” is nothing like this. USS’s annual revenue currently meets its liabilities comfortably, and is projected to do so for decades into the future. (See figure 1)

CashflowFigure 1: USS cash flow to 2068. Source: Salt and Benstead (2017) Progressing the Valuation of USS. London: First Actuarial.

What the USS “deficit” actually measures is something much less determinate, namely any shortfall between “the amount of assets held and the amount of money estimated to be needed to pay the pensions built up.” One problem here, as we and others have stressed elsewhere, is that such estimates depend upon assumptions about demographics, scheme investment decisions, and the future performance of relevant investments. (To get a sense of what is involved here, consider an individual considering how much money they would need to set aside now, and how to invest it, in order to fulfil a commitment to buy a specific house in 2040 at whatever is then the market price.)

But there is another problem, which has been less widely discussed. (Though see Sean Wallis’s excellent blog, and this letter to the Financial Times signed by a long list of statistical experts.) This is the problem of the uncertainty of any estimate of the “deficit” under such a model. Any statement of what the USS deficit “is” is in itself misleading, since such a figure necessarily refers to a point on a probability distribution rather than a bare arithmetic quantity. The real possibilities are in fact spread out around (and mostly above) this point. Figure 2 sketches the distribution for even the “recklessly prudent” November 2017 valuation:

DistributionFigure 2: Sketch of the probability distribution of the USS deficit/surplus under November 2017 assumptions. Source: Sean Wallis, https://heconvention2.wordpress.com/2018/02/08/made-in-westminster/

The sketch makes some key facts clear. First, the expected value of USS assets and liabilities over the long run (the blue line) gives a £5.2 billion surplus, even on the terms of the November valuation. The “£7.5 billion deficit” arises because USS, following the legal requirement for “prudence”, measures the deficit not at the mean, but at the 33rd percentile (the red line). So the figures given for the “deficit“ in reality identify a funding position which will be exceeded 67% of the time. The “deficit” is not only not currently existent, but it is not even something which is likely to happen.

Given this, it is extraordinary that neither UUK nor USS have provided clearer information on the precise degree of uncertainty in their estimates of the scheme deficit. As the FT correspondents wrote in September:

[T]he USS provides insufficient information about the methods used to value its assets and liabilities. They present no confidence intervals around the point estimate of the deficit, indication of estimation error, or any sensitivity analyses. They provide virtually no details of what data or analytic code they used to come to their conclusions. Even the CMI 2014 report underpinning the mortality assumptions is not publicly available.

If the reports from which the USS valuations are drawn were academic papers, the writers conclude, “diligent editors would reject them out of hand”. It is time that both USS and UUK stopped preaching about spurious financial necessity. They need instead to begin to talk honestly not only about the supposed difficulties of funding decent, guaranteed pensions, but also about the real difficulties with their own numbers.