What’s wrong with USS’s claim that short-term risk of 65% investment in growth assets is too high
[Update 9 September 2020: Here’s a brief Twitter thread in which I raise a similar, but simpler, objection to USS’s latest version of its short term risk metric.]
The joint UCU/UUK Valuation Methodology Discussion Forum has asked USS to model “an option involving a higher-return (and higher-risk) investment strategy”, in which the scheme would remain continually invested for the long run in a manner benchmarked against a “reference portfolio” weighted roughly 65% towards “growth assets” (equities and property). This is what is known as the “no de-risking” option, since it would involve no systematic shift of the portfolio over time from growth assets into bonds and other “liability matching” assets. (‘Methodology and risk appetite for the 2020 valuation’, March 2020, pp. 35, 38)
The downside risks of such investment in growth assets would be managed by the setting of a level of employer and member contributions into the scheme that is high enough that the assets purchased by these contributions have a greater than two-in-three chance of delivering investment returns that will cover all pensions liabilities. USS notes that “Setting a contribution rate at [this] higher level than required [by their current approach] enables us to build up a ‘risk buffer’ to mitigate the risk associated with such a higher-return strategy” (ibid., p. 38).
Even when they are set so as to yield a greater than two-in-three chance of delivering sufficient returns, contributions might not need to rise to unaffordable levels. This is because the expected returns on growth assets are so much higher than the expected returns on bonds and other liability-matching assets into which the scheme would otherwise de-risk.
Moreover, USS acknowledges “that the ‘risk buffer’ built up over the next 20 years could provide effective risk mitigation by 2040 with sufficiently high contributions” (ibid., p. 45). In other words, what they describe as the “long-term risk” of such an approach might be acceptable.
USS expresses concern, however, over what they describe as the “short-term risk” of such an approach: “as it takes time to build up the buffer, [this approach] does not address the higher risk over the short term” (ibid., p. 38).
USS’s measure of short-term risk is the gap between (1a) the current value of the assets and (1b) the higher cost of now purchasing a gilts+0.75% self-sufficiency portfolio plus (2) the increase in this gap over the next year that has a 1 in 20 chance of arising in a worse case scenario (ibid., p. 51).
If this measure exceeds the “risk appetite” of employers by too great a margin, USS deems short term risk too high and maintains that steps must be taken in response.
USS defines the employers’ “risk appetite” as the value of the extra contributions they would be willing to make over a large number of years in an adverse scenario, in order to eventually raise enough money to purchase a gilts+0.75% self-sufficiency portfolio by the end of that time period. USS works from the assumption of a risk appetite of +10% contributions over 30 years, the present value of which is £35bn. (Ibid., pp. 26, 29, 51)
The following question arises: Since the journey to the “safe harbour” of a low-risk self-sufficiency portfolio would be incremental and not be complete for another 30 years, what is the relevance of today’s self-sufficiency deficit, and the 1 in 20 risk of an increase in that deficit over the next 12 months, to the feasibility of such a long and incremental journey?
In other words, why should it matter how expensive it might become to purchase a self-sufficiency portfolio today, or within a year, given that it would make no sense to purchase one so soon even if it were possible to shift the assets so swiftly? As USS sensibly maintains, they “certainly do not expect to have to move to a self-sufficiency strategy in the short term” (‘2018 Actuarial Valuation’, January 2019, p. 12).
Given the “safe harbour” rationale USS has offered for self-sufficiency as a means of securing of pensions promises (see ‘Methodology and risk appetite’, p. 26), only the following question regarding the long term is relevant:
Q: Would it be possible, via +10% of payroll each year for the next 30 years, to purchase a gilts+0.75% portfolio by year 30?
There are broadly two different methods of answering this question.
The first method, preferred by financial economists, is grounded in market prices.
The second method relies on assumptions and forecasts that depart from market prices. USS’s forecast regarding gilt yields in 20 years’ time, for example, assumes gilt yield reversion out of line with “market breakeven levels (forwards)” (‘2018 Actuarial Valuation’, p. 16).
This second method involves a relatively high degree of subjective judgment and model risk. It may therefore also be useful for USS to perform cross-checks involving the more objective, market-based metrics that financial economists appeal to.
But the contrast which USS needs to draw is not between current or short-term risk and long-term risk. Given their 30 year journey rationale for the significance of self-sufficiency, USS must instead draw a different contrast: between a more objective market-based method, versus a more subjective method involving greater actuarial judgment, of estimating the cost of purchasing a gilts+0.75% self-sufficiency portfolio in 30 years’ time.
The relevant market rather than judgment-based metric is the market-derived price (involving forward markets) of purchasing a gilts+0.75% portfolio in 30 years’ time, rather than the market price of purchasing a gilts+0.75% portfolio today. USS has not provided a sound rationale for its focus on the later rather than the former.
To be time-consistent, USS’s construction of the relevant self-sufficiency deficit will also need to compare the expected level of the scheme’s assets in 30 years’ time with the market-derived cost of purchasing a self-sufficiency portfolio in 30 years’ time. To calculate the level of the assets in 30 years’ time, USS might apply a more market-based forecast of expected returns on scheme assets over the next 30 years, rather than their “fundamental building blocks” approach which currently assumes gilt yield reversion.
Inevitably, any such long-term forecast will need to involve a fair amount of actuarial judgment. It will be much more subjective and less clear than an estimate of the market value of the scheme’s assets today, which is used to calculate the current self-sufficiency deficit. But if one were to maintain that we should use the latter metric because the value of the assets today is much clearer than the value of the assets in 30 years’ time, that would be akin to a drunk’s saying that he should search for his keys here under the lamppost because things are much clearer here than they are in the darkness 30 yards away, where he dropped his keys — where I take 30 yards to be an analogy for 30 years.
In sum: USS has not yet provided a justification for its focus on the current self-sufficiency deficit — where that is the current market value of the assets minus the current cost of purchasing a self-sufficiency portfolio — and on how that deficit might change over the next 12 months.