Wilkinson and Curtiss: Current discount methodology is an elephant in the room
Controversy over the discount rate used to value defined benefit pension liabilities is nothing new but, as Tim Wilkinson and Frank Curtiss explain, the flaws may be more serious than many realise.
The creation of an expert panel to review the valuation of the Universities Superannuation Scheme (USS) has kept the scheme open, at least for the time being. But for defined benefit (DB) provision generally, the picture is grim. No FTSE 100 company now has a DB scheme open to newcomers. BT and British Airways have recently closed large schemes citing, as did the USS before its reprieve, the impact interest rates have had on deficits.
Discount rate controversy is nothing new. One rarely, if ever, hears people in the industry say that using the yield on high-quality corporate bonds (as accountants do), or a rate just above gilt yields (as most actuarial valuations do) is without problems. But the flaws are more serious than many realise. The theoretical case for these rates is acutely defective. They have wrecked company balance sheets, caused the misallocation of billions of pounds of corporate resources to plug illusory deficits, distorted scheme investment strategies, and played a major part in the collapse of private DB provision. If a disaster even a fraction of the size had befallen the state pension system, governments would have been voted out of office. It's a national scandal.
The rot set in in November 2000, with the publication of the accounting standard FRS 17, the spirit of which is still with us today in the form of IAS 19 and FRS 102. Both these standards require liabilities to be discounted by reference to ‘high-quality corporate bonds'.
Plenty of people spotted there was a problem. The Confederation of British Industry (CBI) said FRS 17 would undermine final salary schemes, while Peter Thompson, who was chairman of what is now the Pensions and Lifetime Savings Association prophesied, correctly, that it would "drive many employers from providing defined benefit pensions".
Sir David Tweedie, the chairman of the Accounting Standards Board (ASB) while the standard was being developed, acknowledged the discount rate was an issue. Interviewed in 2006, he said the corporate bond rate was chosen because: "Everyone else around the world used it. We had given up by that stage; we had spent two years trying to find a discount rate."
It's not surprising that the ASB had difficulty finding a suitable rate. In reducing an entire pension scheme to two numbers, too much information has to be thrown away. The best tool for monitoring a pension scheme is a cashflow forecast. The conversion from forecast to balance sheet has to be done, but it should be done with due respect to the principle of neutrality. If a scheme has healthy cashflows in perpetuity, it is simply bad accounting to do the arithmetic in such a way as to produce an illusory deficit.
With deficits being merely an artefact of measurement, rather than a reflection of underlying economic reality, the unfortunate outcome was poor decision-making. Some schemes altered their asset mix in an attempt to avoid volatility or hedge liabilities, turning away from investments that would almost certainly have given better returns in the long run. The resulting demand for a finite supply of bonds drove down yields, and hence discount rates, in a vicious circle that increased the (already overstated) liabilities. Even Tweedie described this result as "daft".
The time value of money is an entity-specific, or even a project-specific, concept. It only makes proper sense when chosen relative to the time preferences of the entity concerned, and after due consideration of the alternative uses to which the money could reasonably be put. The most suitable choice in the case of a pension scheme is therefore the best estimate of its future investment return; and this also produces the result that corresponds most closely to a cashflow forecast. If a scheme is required to pay £100 in a year's time, and only has £91 today, it will still be able to meet its liability if it has an investment return of 10%. Divide the £100 liability by 110% and you get £91, so the scheme is fully funded.
A discount rate based on investment return was considered by the ASB, but rejected8. In fairness, there are complications. Future returns have to be estimated, and this introduces the opportunity for manipulation. Schemes may also be tempted to take on too much risk in order to suppress liabilities - the opposite error from over-investing in bonds. These are reasonable objections, but ultimately they fail. Society would have been better served by having auditors and regulators spend their time policing over-exuberant return assumptions than from having them preside over death by discount rate.
A factor commonly cited in support of IAS 19 is consistency - all companies have to use a similar rate. Because schemes do not all follow similar investment strategies, however, the additional cash they will need from their employers in the future will differ widely, even for those with the same deficit. Therefore, the IAS 19 rate is also inconsistent with the concept of a true and fair view, since IAS 19 liabilities do not legitimately represent a real future obligation to pay.
Another productive way of looking at the problem is to work backwards (as did First Actuarial's Hilary Salt and Derek Benstead in their September 2017 report for the University and College Union (UCU), Progressing the valuation of the USS) - asking questions such as ‘for past service, at what break-even discount rate will the liabilities have the same value as the assets?' and ‘for future service, at what discount rate will future inflows have the same present value as future service outflows?'. Though less suited to financial statements, these calculations can be an extremely useful tool for planning. If the break-even rates are less than the likely rate of future investment returns, there is no need for additional payments into the scheme.
Turning to actuarial valuations, the scheme funding regulations are superior to accounting standards, since they allow the rate to reflect the return on scheme investments. In practice however, according to data from The Pensions Regulator (TPR), the median scheme uses a rate of about 1.1 percentage points above 20-year gilts.
There are a number of reasons for this. The scheme funding regulations require assumptions to be chosen prudently. The regulator is also of the view that the rate should be lowered where the employer covenant is weak. The logic of the latter policy, however, is highly dubious. Boiling a pension scheme down to two numbers already means asking the discounting calculation to do more heavy lifting than it is really capable of. The starting point for a discussion on deficit correction should be a cashflow forecast, not a shortfall inflated by the risk of employer failure.
The result is that trustees, under pressure from TPR, have settled into overly prudent practices. A long run return one percentage point above the 20-year gilt rate, which is itself hovering just above historic lows due to quantitative easing, is completely at odds with what trustees believe when they are devising investment strategies, appointing investment managers, and setting benchmarks. For example, despite the bursting of the dotcom bubble in 2000, and the financial crisis of 2008, the total return on the FTSE All Share Index from January 2000 to April 2017 was 4.6% per annum.
Returning to the USS, according to its reports and accounts for the year ended 31 March 2017 it has achieved a return of 12% per annum over the last five years, while the total value of the fund, net of contributions and benefits, increased from £34.2bn to £60.5bn between 2012 and 2017.
Yet here - almost unbelievably - is an extract from the 2017 report and accounts: "…the deficit on the technical provisions basis … has increased from £5.3bn in 2014 to £12.6bn at 31 March 2017. The investment performance … has not outweighed the effect of the fall in discount rates which has led to the liabilities increasing at a faster rate … over the period."
There is an elephant in the room. What we are witnessing is not prudence; it is prudence gone mad.
First Actuarial, on behalf of the UCU, has developed a USS cashflow forecast, and also calculated the scheme's break-even discount rates of the kind alluded to earlier. The results are striking. Despite its supposed £12.6bn deficit, at current contribution rates the scheme can pay benefits until at least 2068 with virtually zero reliance on either capital gains or investment income. Furthermore, the break-even discount rates for both past and future service are well below the expected returns on equities and property. On any reasonable calculation, the USS is in perfect health. The biggest risks it faces are bad accounting and poor regulation.
Fortunately, there are signs that the industry might be receptive to change. Michael O'Higgins, who was chairman of TPR from 2011 to 2014, is open to the idea that cashflow forecasts might be a way forward, writing in Professional Pensions that they make it "much plainer what incremental investment return would be needed to close any cash flow gaps, and trustees and employers could together consider whether any additional investment risk to achieve that increment was worth taking, or whether additional contributions would be preferable".
Perhaps the elephant is visible after all. But changes to the accounting and regulatory framework need to be made quickly, while there is still something left to save.
Frank Curtiss is the immediate past president of the ICSA and the former head of corporate governance at RPMI Railpen. Tim Wilkinson is the former chief accountant at RPMI Railpen
