Seatbelts and Brakes
An earlier article (Of Dervishes and Mendicants) illustrated the surprising fact that the application, since 2000, of a further £340 billion of funding to UK defined benefit pension funds, more than 40% of reported liabilities in the year 2000, has not improved their condition. It also showed that contrary to expectations radically revised asset allocation and more advanced management techniques such as liability driven investment had not been effective. This article will consider possible explanations.
The start point is a comparison of funding or contribution cost payments with increases in the actual pensions ultimately payable. Figure 1 shows the total contributions made (in constant 1992 pounds). It also illustrates a real pensions cost index which is derived from a factor model, which utilises the determinants of final salary pensions, such as longevity and wage increases; interest rates are not a parameter of this model. The figure also shows a present value index, which reflects the effect of changing AA corporate bond yields on the factor cost index.
Over the nineteen year period shown, the real cost of a final salary DB pension has risen by some fifty percent due principally to increasing longevity and earnings. The funding or contribution cost tracked this increase rather well until 2002, but thereafter departed violently from it.
By contrast, as interest rates fell, the present value cost index departed markedly from both contributions and real pension factor costs. From 2002 on, it is evident that the present value is the driver of actual cash contributions made by sponsors, rather than the ultimate cost. Though FRS17, the UK accounting standard, which specifies the use of a AA corporate bond rate in calculation of pension scheme liabilities, was not mandatory until 2003, it was widely discussed following the release of the exposure draft in 2000 and early adoption was encouraged. This is a clear case of the accounting influencing actions, rather than merely reflecting them.
Given this change in accounting standard in 2003, we next examine, in Figure 2, the evolution of scheme liabilities as reported by the Pension Protection Fund and compare those with the factor cost index and the underlying PPF index adjusted for the effects of changing discount rates.
In the period since the introduction of the Pensions Act 2004, the reported liabilities of schemes have risen more rapidly than the factor costs of final salary pensions. Even if we dismiss the aberrant behaviour of 2008 as being an exceptional situation, it is clear that rather large differences have emerged. These are all the more surprising when scheme activity over this period has been highly focussed on reducing pension liabilities. Enhanced transfer offers, pension income exchanges, closure to new member shifts to career average, and closure to future accrual have all taken place. Some on a very substantial scale – the Association of Consulting Actuaries recently reported the results of a survey of 650 schemes which reported 91% closure to new members and 37% to future accrual. Though “buy-out”, the winding-up of a scheme through annuitisation with an insurance company, has been the subject of much press commentary, its scale has been small, but it would also tend to reduce the total stated PPF liabilities.
The divergence is very substantial; at 2010 of the order of £110-£120 billion, around ten percent of total liabilities and more than 8% of GDP. It appears that the advent of the PPF and pension regulator have changed the way in which trustees derive their valuations. We already know that regulation has had a pronounced effect on contribution rates, which are shown in Figure 3. Both special contributions and ordinary contributions have been rising at rates far higher than their historic. Special contributions, which are dominated by deficit repair schedules, are almost entirely and explicitly induced by regulation, and related to the present value rather than the factor cost. It appears that a material part of the increase in ordinary contributions may also be attributed to the effects of regulation.
The puzzles of lack of improvement in funding status and failure of liability driven investment to perform as expected may be decomposed into issues related to the factor cost of pension provision and the accounting issue of the discount rate applied to reduce those ultimate liabilities to a present value.
As the accounting values are directly related to contribution cost, there has been a marked trend for schemes to adopt liability driven investment. The first step in such techniques is the hedging of discount rates either using interest rate derivatives or increasing fund allocations to long-dated bonds. This is actually perverse behaviour as it can be demonstrated empirically that corporate profitability, and their ability to pay pensions increases, and with that member security, is high when interest rates decline or are low. These relations are also expected theoretically. Only recently have we seen publications, such as KPMG’s Pension Repayment Monitor, which take notice of the strength of the UK corporate sector and the increase in member security which results. It is to be expected that corporate profits will be high when government deficits are large. Figure 4 exhibits the relation between the level of interest rates and the profits of private non-financial companies. It should also be noted that the profits used in this figure are low-biased; they are reported after payment of pension contributions. Put another way, the scheme exposure is a natural hedge of the exposures to interest rates of the sponsor. Rather than “de-risking” the company, hedging the scheme interest rate exposure actually increases its effective gearing.
Incidentally, this natural hedge characteristic extends over many other pension risk factors, such as inflation and longevity. There are actually very few companies which do not benefit from the increased demand of a larger population arising from increasing longevity.
Liability driven investment is the attempt to increase the co-variance of pension fund assets and scheme liabilities. The first step in this process is usually the hedging of interest rate exposures using derivatives or long bonds of conventional form. In addition to this, as practised, liability driven investment usually involves the hedging of inflation risk, as scheme liabilities are linked to inflation.
It is worth considering the position of the company in this context, since this is the entity which is exposed to these unconditional scheme risks unless insolvent. The relation between company profits and interest rates is negative; lower interest rates are associated with higher corporate profits. There is a natural offset of the interest rate risk between scheme and sponsor.
Similarly, there is a relation between inflation and corporate profits; this is also negative. High inflation and pension cost is associated with low corporate profits. Consequently it may make sense for the company to hedge inflation risk and effectively de-gear itself with respect to this risk factor.
In fact, as interest rates and inflation are themselves related (in technical terms multi-collinear), it is necessary to hedge interest rates and inflation jointly using an estimation technique, such as partial least squares, which is robust to this property.
However, schemes are not exposed to inflation in its broad sense; they are only exposed to limited price inflation. The relation between limited price inflation and corporate profits is markedly different from general inflation as can be seen in Figure B1.
A simple model using only gilt yields as the explanatory variable accounts for 78% of the variability of corporate profits. When we consider both gilt yields and inflation, again in a linear model, the explanatory power of the model increases only marginally. It is notable, though, that the weighting on interest rates declines markedly.
When we consider a linear model which uses both gilt yields and limited price inflation, the explanatory power declines significantly. However, the surprise here is the sign of the weighting on limited price inflation. If it was correct to hedge inflation by buying securities which deliver inflation-linked returns to the scheme, then, in this joint context, it is correct for the scheme to hedge limited price inflation by creating inflation linked securities which pay limited price inflation to their owner.
It appears that those who sold their holdings of inflation-linked securities in recent times – more than 25% of schemes (see Figure B2 below) – may have been thinking about rather more than their relative expensiveness.
The central problem here is that both the regulation and the hedging consider the scheme risk as if it is an unconditional risk. The motivation, though, is protection of pension-holders, for whom this risk is, in fact, conditional. Until the sponsor is insolvent, the level of scheme funding is immaterial to a scheme member. These risks are unconditional risks to the sponsor, but then any hedging should be done in the context of the sponsor company’s aggregate exposure to the risk factor. This would involve also consideration of the relative sizes of scheme and sponsor.
We should not be surprised that regulators operate on scheme funding, since the primary risk, sponsor insolvency, is not available to them as a control variable. In western free-market economies, it is almost an article of faith that government should not intervene to manage the insolvency risk of private sector companies – such management is a private sector concern. This leaves only the secondary and conditional aspect of risk, funding and asset allocation available to them. It is regrettable that the authorities sometimes refer to their asset-based operations as management of sponsor insolvency risk. It is, of course, inefficient to operate on the conditional element of the risk and can result in manifest paradoxes, such as consequential increases in the primary risk, the sponsor insolvency likelihood.
Regulation lays claim to having increased member security. It is worth considering the counter-factual. If instead of funding their schemes, the private sector had reinvested this in their commercial activities, by how much would their insolvency likelihood have declined? The answer to that is substantially – it is over 40% of capital employed in the year 2000. And how would the economy have performed in those circumstances?
Yet at a recent conference, we heard Alan Rubenstein, the CEO of the Pension Protection Fund, utter the battle-cry “Funding trumps Covenant”. This is complete nonsense. It is as ridiculous as the statement: “