Iain Clacher & Con Keating
As part of our Funding Code research, we searched for academic or practitioner papers covering long-term expected returns forecasts. We were particularly interested in the ex post accuracy of these forecasts. We found none which used historic market performance[i] other than for short-term concerns such as corrections to market bubbles and periods of boom and bust. There were a few, macro-economic in nature, where long-term returns are functions of growth and demographics. To use an analogy, this is climatology rather than weather forecasting. We shall revert to this later.
What we do know
There are a few things we do know about gilt yields – they are strongly predictive of future long-term gilt returns, but that relation is tautological. They are not predictive of equity, property, or other asset class returns at any holding period horizon. This renders their use in gilts + presentation of expected returns highly questionable. To misquote Ralph Nader, they are unsafe at any horizon.[ii]
We have spent much of the past week trying to reconcile various claims and figures cited in the latest USS valuation consultation with UUK as these two things seem inextricably linked. We have had little success.
As best we can tell, the single equivalent discount rate for USS would be less than 2% nominal and the required rate of return on scheme assets would be around 3.2% nominal. These seem to us to be low and readily achievable. With that in mind, we looked to the long-term expected returns forecasts of other long-term financial institutions. The expected returns of other UK pension funds are not a valid comparator as they are subject to the same regulatory panopticon.
Looking further afield, The Norwegian Fund for Future Generations publishes its expected returns – they expect 3% real above their CPI which has averaged around 1.75% in recent decades, so a nominal of around 4.75%. The risk (volatility) of their portfolio is 12%. The most interesting aspect of all of this is that in 2017, in response to declining government bonds yields globally, they moved their target asset allocation from 60/40 equity/bonds to 70/30 equity/bonds and increased the expected return to 3% real from 2.75%. In the context of this shift in investment strategy and return expectations, it is worth bearing in mind this is the largest of the sovereign wealth funds (with circa $1.2 trillion of investments as of July 2020)[iii]. This is a fund that has access to the best advice in the world. Moreover, it has achieved these types of returns over the long-term[iv].
By our calculation, if USS were to use this rate, it would not be reporting any deficit but rather a surplus of similar order to the headline-grabbing £18 billion deficit.
This issue of what do future returns look like (and what returns do we need) is also linked to the recent blogs on cash equivalent transfer values.
If we fund a scheme to the levels of liabilities arising from low rates of interest, we are effectively pre-funding those liabilities relative to their contractual values. This also has the effect of lowering the required rate of return on the asset portfolio, and with that, the potential future cost to the sponsor employer. If a scheme is fully funded at this rate, the required rate of return on assets is that rate.
In these circumstances, if a member takes a transfer based on these values, albeit that the transfer may be limited to the degree of funding of the scheme, then it is crystalizing the employer’s cost to that date. Crucially, this transfer enables the employee to extract all of the pre-funding, and denys the employer the possibility of recouping the costs of this prepayment, as any future outperformance of the asset portfolio relative to this low return, is no longer possible on the assets that have been removed from the scheme.
While it may be the case that transfer transactions throw up gains in accounting terms when those accounting liability values are inflated by the use of low gilt rates, but that is a short-term accounting gain, which comes at the expense of longer-term real gains from higher returns than those we currently observe in the market.
Moving to a gilts-based de-risking strategy has the same effect of crystalising the elevated sponsor costs while removing any possibility of recouping them.
It is worth comparing these transfers with the size of the PPF; at £80 billion they are four time the liabilities of the PPF, and in reality, the PPF is rather small relative to the overall DB pensions marketplace. In its 15 years of existence, it has assisted just 2% of scheme members and less than 1.5% if measured by liabilities, and it has done so at eye-watering cost. It is apparent that the fear of sponsor insolvency greatly exaggerates the actuality.
The Funding Code consultation makes much of protecting members. This raises the question of just how much of members’ pensions is at risk. The answer is rather little. If we consider the scheme we used in illustrations in earlier blogs, we have 63% as pensioners in payment and 37% deferred. The pensioners in payment are fully covered by the PPF so the risk exposure is limited to the 10% haircut applied by the PPF – so just 3.7% of future pension payments. These have a total future value of just £756k – a small fraction of even the minimal present-day funding cost of the proposed code. It is not difficult to conclude that this funding code strategy is more about protecting the PPF than members.
A final thought
The ‘lower for longer’ view of interest rates is now conventional wisdom. As such, it and its associated returns expectations are suspect. The shifting global demographics imply that we are moving over the coming three decades from the deflationary environment of the past three decades to one in which inflation and higher interest rates will prevail and with that low growth. This unconventional view is explored fully and coherently in Charles Goodhart’s latest book, ‘The Great Demographic Reversal’; we recommend reading it. This is a change in the financial climate that is perhaps as important as the change in the natural climate.
One consequence of that would be that this is surely the wrong time to de-risk in the manner proposed in the DB Funding Code.
In the brief time since we wrote this blog, we have had some very productive discussions with some of our peers. Our attention was drawn to the Canada Pension Plan which publishes 75 year return expectations for each of its two funds. These are 5.95 % (CPI + 3.95%) for the ‘Base’ fund and 5.38% (CPI + 3.38%) for the more conservatively allocated ‘Additional’ fund. Obviously, it is too soon to evaluate the accuracy of these forecasts but the indications to date are supportive. It is notable that both CPP and the Norwegian employ peer review of their assumptions. We feel that the Pensions Regulator’s prescriptions should be subject to similar peer review.
It has also been pointed out to us that the large Canadian funds have proved able to harvest ‘illiquity premiums’ very successfully, with which we agree. However, we will make just one point here, though we will return to the subject in our commentary on the proposed second Code consultation. That point is that it is liquidity in the sense of tradability which has a cost rather liquidity which receives some extra compensation. The means that if you buy liquid securities you pay this cost regardless of whether you exercise the option to use it by selling in a market. Gilts, of course, are the most liquid and most expensive of securities from this perspective. One of the effects of quantitative easing is to lower the cost of liquidity, though relative value differences should persist between on and off the run securities should persist, This lowering of the cost of liquidity should also result in a greater reluctance by dealers to hold large inventories of bonds in pursuit of their liquidity provision role – the returns to capital are less attractive.
Finally, we have had much commentary on the prudence of buying gilts at times when their expected returns are negative in real terms[v]. However, as we have been promised a definition of prudence by the Regulator in the second consultation, we shall leave further discussion until that point in time.
[i] The long-term memory literature results for UK markets are mixed.
[ii] The original comes from Unsafe at Any Speed, Ralph Nader, 1965.
[iv] Those interested in more detail should read their White Paper (in Norwegian) available at:
[v] For more on this aspect see: https://www.bankofengland.co.uk/statistics/yield-curves