The Time Line
In pensions is all-important. The need for a pension arises in the first place from the asynchronicity of labour income and consumption. A pension is simply an income in retirement; it is not savings for retirement, which will require realisation to generate income. It is notable that individual DC plans fail this most elementary test; they are tax-advantaged retirement savings schemes and should not properly be known as pensions in the absence of some defined life annuity conversion capability.
Economically a pension is a claim on future production. The logic of occupational pension provision is precisely that it is these employers, together with their heirs and successors, who will be generating that future production, just as they are producing today. These are the institutions which can credibly contract on future production, since this is their existential motivation. Indeed, it is this contractual credibility which allows them to issue debt and equity securities, many of which are traded in capital markets. These securities are also claims on future production.
In fact, the existence of such traded claims allows other institutions to be devised which may offer pensions; usually, these take the form of insurance companies. The ability of an insurance company to offer pensions depends upon its ownership of such claims on production, its assets. In this the insurance company is operating in a redistributive role; it is an inside institution while the household claim arising from employment is outside. The inside institution is dependent upon financial markets and the relation of financial markets to the real economy, while the outside household occupational claim is direct and dependent upon the market for the goods and services produced by the employer. The employer may have secondary dependence on financial markets. The indirect financial institution route is less efficient. Notwithstanding the elementary economic theories which link financial markets to productive output, the empirical evidence is that these are largely unrelated.
Figure 1 shows the returns to capital of the UK private sector and the annual total returns of the FTSE All-Share index to illustrate this point. Over this period the correlation was negative and the explanatory power of the linear regression statistically indistinguishable from zero. It is also notable that the volatility of financial market returns is an order of magnitude greater than the volatility of PNFC earnings. There is a consequence to this relative volatility. For any given level of member pension security chosen, the level of capital held would need to be far higher for the insurance entity than for the private sector producer. Another way of saying this is that the guarantee offered under UK DB arrangements by non-financial sponsor employers is far more efficient than such a guarantee emanating from the financial sector. This is a substantial rebuttal of the European insurance and reinsurance federation (CEA) calls for equal treatment of pension schemes and insurer under Solvency II.
This is not to say that inside arrangements, such as insurance, are entirely redundant; there are many, for example, the self-employed, who must rely upon either pensions organised in this manner or other sources of retirement income, such as residential property.
In a recent conversation on pensions with the retired former finance director of a major UK company, he observed: “When I set up our DB scheme, I committed to paying pensions to our retired employees as they came due; this was not, and was unlikely to become, a meaningful burden for the business to meet from its current and ongoing activities. Public policy was to encourage occupational pensions.” This was a question of sensible industrial relations. He continued: “In setting up a trust to administer the scheme, I accepted the position of an unlimited liability shareholder: if the trust had insufficient funds, our business would have to pay, and if all pension liabilities had been discharged the business could recover any residual assets.” Unlimited liability is slightly misleading inasmuch as the liabilities incurred are finite in amount . However, it also makes evident a particular aspect of pension trusts – the sponsor maintains an interest in the fund even though that exists for the benefit and security of pensioners, past, present and future.
If regulation induces funding in excess of that which proves necessary for the discharge of liabilities, this is the property of the sponsor and reverts to them. This would imply a cost to the sponsor which is related to the interim unavailability of these assets rather than complete loss of them – a much lower cost usually. However, it also makes it all the more important that regulation does not induce behaviour which incurs other costs and expenses; expenses do seem to possess the Parkinson property of expanding to consume the funds available.
However, the most important aspect here is the pension payment cash-flow view. Of the two insolvency doctrines, equitable insolvency and balance sheet insolvency, this is the equitable view. This view considers the default on a payment when due to constitute insolvency, while the balance-sheet view considers an excess of liabilities over assets to be sufficient. The accounting and regulation in force are of the balance sheet type. Most of modern academic financial analysis is conducted under a balance sheet view, for example the 1974 Merton model for the evaluation of corporate credit. The consequences of this difference are discussed later, but readers are also referred to the UK’s 2011 Whole of Government Accounts which contains a discussion of these issues in a national accounts context.
To summarise some of his other points: Many of the terms of the pension award were products of the time. Final salary required the capture and maintenance of just one record; it was administratively convenient and probably also important in recognition of the lack of inflation indexation of pensions in payment at that time. Poor transfer and preservation terms were recognition that most of the early leavers were going to jobs with higher compensation packages; jobs that they were able to secure in part because of the sponsor’s investment in their human capital.
“I agreed to funding of the scheme as this could be used to smooth demands upon the finances of the business, and would be particularly useful in difficult times. The investment income of the fund and employees’ contributions were important in this regard. You also need to understand that the ability to roll up investment income and capital gains within a scheme without incurring tax costs, at that time, was worth a lot – it reached levels as high as 5% or 6% a year for our higher paid staff in the 1970s. Our pensions booklet made much of the benefit for employees to saving under such a tax-advantaged arrangement.”
Today’s discussions of the purpose of scheme funding attribute very different motivations to scheme funding and centre on the fund as collateralisation of the sponsor promise and the production of sufficient income returns to meet the pension promise independently of the status of the sponsor employer. Interestingly recent surveys of pension scheme managers report that they do not now consider the income characteristics of their asset allocations, but rather total return. The income yield, at market prices, of UK pension schemes was below 2% in 2010. This is problematic in two regards. In the short-term pensions payable are projected to rise further, requiring either more special contributions or the sale of assets to meet these payments, and of course, few will be permitted by the Pensions Regulator to sell assets given the magnitude of scheme deficits. In the long term, realised investment returns are dominated by the income yield; in their classic 111 year study, Dimson, Marsh and Staunton attribute 4.8% of the total real return to equities of 5.5% to dividend income and a further 0.5% to dividend growth. The long-term problem would seem to imply further radical revisions to asset allocations are necessary.
The former finance director continued: “What I did not commit to was the payment to the scheme of capital sums sufficient to meet all future pension payment obligations. Nor did I commit to making payments sufficient to maintain such a full funding status. Those would have been complete folly.” His explanations of why such practices are “pure folly” are illuminating. “I was happy to make contributions which equalled the wages forgone by employees for current service, so that there might be no direct financial gain to the business in its current activities, because the pension scheme helped to deliver a more stable workforce and better industrial relations. In turn that meant that we could invest more confidently in our labour productivity. However, I would not want to leave the impression that the cost to the business was actuarially fair; without the tax concessions we could not have afforded, for the contribution we made, the benefits offered without the roll-up tax concession. The price of this was that we would be liable to the scheme for our past mistakes, our errors in the assumptions made about the investment returns to be expected from these assets. Make no mistake, this was always the foremost issue – we calculated that 85% or more of the employee’s pension came from investment returns, not the contributions. Our professional advisors were of little use in this.” Note that there were incentives for the sponsor employer arising from the tax-deferments in pension funding; moreover it should be realised that as both corporate and personal tax rates have fallen, so have the values of these concessions as incentives.
There is an interesting calculation to be done. If the value of the tax concession is less than the cost of regulation of DB schemes, it makes no sense to offer such qualifying regulated schemes. With the value of this tax concession now worth less than 2% p.a. even to a high earner, it would appear that this point was passed long ago in the UK.
One of the issues currently absent from the DC versus DB debate is the cost of these tax deferments; with DC arrangements, depending upon the precise institutional design, having an efficiency of between 45% and 70% of collective occupational DB for the provision of a unit of pension, the incurred tax costs differ markedly. The provision of DC by an employer is also explicit abrogation of their role as a direct supplier of the employee-required claim on future production; these employees are entirely dependent upon the financial markets in these claims.
The conversation continued: “The real problem with funding regulations, as they are currently implemented, is that they import the cycles and madness of financial markets into the business’s real financial activity, and we should not forget that all businesses are faced by the discipline of the markets for their goods and services.” The current debate over quantitative easing is a prime example of such financial market effects. If we suppose that the Bank of England’s recent actions lower the financing costs of the UK private non-financial sector by 0.5%, we see the direct stimulus of a lowering of £4 billion in their borrowing costs, but the effect of this lowering of discount rates is to increase their pension scheme deficits by £40 – £50 billion. In the absence of forbearance by the Pension Regulator, this is likely to generate additional funding demands of similar magnitude to the direct stimulus under ten-year deficit repair programmes. This would not entirely negate the stimulus through this channel, but the effects would now be predominantly in cash-flow timing differences.
It is also clear that these volatility effects and the resultant demands for additional funding have been material drivers of the various practices known as “de-risking”, such as liability driven investment . We shall confine ourselves, here, to the simple observation that hedging is a form of speculation, driven by market price movement rather than the productive returns of investment. Of course, this distinction is principally one of time or investment horizon; long-term realised total equity returns result principally from dividend income received and to a far lower extent on dividend growth, while short-term returns result principally from change in the market price. The balance sheet view induces such hedging and short-term speculation.
In response to the question as to how the investment return assumptions of this former finance director’s scheme were set, the answer was: “We developed simple estimates of future returns in consultation with our actuaries and fund managers. It was all ad-hoc, guesswork, you might say. By comparison with our one and three year business forecasting and budgeting, it was ‘toy-town’ in nature, but it was also unimportant. You have to remember that our contribution rate was a fixed proportion of wages and widely advertised in the pension materials we gave to employees. This was the figure that the unions focussed upon; they saw this as the wage substitute. That contribution was budgeted for and paid every year. The actuarial analysis was there simply to reassure us that we weren’t hopelessly off-track with our contribution rate, and that revision of it, or other pension terms, or special contributions were not advisable. The returns projections were usually above our own forecast returns to our capital, but in the early/mid 1970s these estimates were below our cost of bank borrowing, so we substituted that as the expected return. Under the lower expected return, a higher contribution rate was recommended. We reasoned that to pay higher contributions based on the lower expected return would hurt the business and that we might as well make future payments to the scheme as to the bank, so we did not increase the contribution rate. There was, in fact, just one year in which we made a special contribution – a year or two after that decision not to increase our contribution rate. The effective rate, on what was in essence borrowing from the scheme, was far below our bank borrowing cost but considered sufficient by the actuary. Markets had also by then recovered from their lows.” This is a stark contrast to today’s world and indeed that proposed under Solvency II. There was no mention at any point in this conversation of the possibility of sponsor insolvency.
The central and genuine problem of corporate finance which lies at the heart of the issues of pension funding is precisely that prior to insolvency the pension scheme does not have a claim for more than 100% of liabilities – in fact, its claim is limited to the difference between 100% of liabilities and any assets which it may hold. This is a matter of equity for other stakeholders, notably other creditors. However, on sponsor insolvency, even if funded at 100% of the best estimate of liabilities, the scheme, if left to stand alone and run off liabilities, has a fifty percent chance of being unable to do this and itself failing. This level of insecurity is considered socially undesirable.
Insurance companies face this intrinsic problem in their activities and the answer which has been developed for them is to require additional capital buffers sufficient to lower their insolvency likelihood to a one in two hundred years event, 0.5%. There is some confusion in terminology here as a 0.5% annual insolvency likelihood means that the insurance company has a fifty percent chance of failing, i.e. is expected to fail, within one hundred years. However, to reduce the likelihood of a stand-alone pension scheme failing due to the uncertainties it faces, the amount of additional funding is calculated by reference to the Section 75 value of the Pensions Act 1995. This funding level is also known as the ‘buy-out’ value, the cost of transferring the scheme to an insurance company. This “buy-out” process terminates the ongoing provision of new pensions. This value varies with market conditions and the precise detail of the scheme but typically lies in the range 130% – 150% of the best estimate of scheme liabilities. Note that the stand-alone scheme faces the full volatility of markets and is indistinguishable from an insurance company in this regard. The central point here is that funding is at best an incomplete security mechanism and one which brings new issues to the problem.
The ASW pension case is interesting inasmuch as it was the prior regulatory attempt to deal with this issue, by altering the priority in insolvency among members of a scheme, to reallocate the distribution of assets among member claims, which was the root of the problem. Pensioners in payment got priority and active members could lose all – the result was naked protesting pensioners on Brighton beach.
Formally, insolvency is the inability to pay one’s debts in full as they come due, but in practice may be characterised in a number of ways. The failure to meet obligations as they fall due is known as equitable insolvency, while the condition that stated liabilities exceed assets is known as balance sheet insolvency. In the former a default has occurred, while in the latter it has not. It is frankly doubtful that any pensioner would be able to obtain a Court judgement based upon the argument that the scheme would likely prove to be unable to meet a pension payment fifty or more years in the future. It is notable that corporate bond indentures require the occurrence of an event of default to have occurred, and not to have been cured or remedied within an allowed period, before the bond-holder owner can accelerate, or move to the present the claim on the company. This is equitable insolvency.
There differences between these two doctrines lie principally with future income and expense. The equitable balance sheet of a pension scheme would include the present values of future expenses and future contributions, and is sometimes referred to as the comprehensive balance sheet; this is the inter-temporal budget constraint. The ordinary balance sheet approach does not include these items; this is the conventional financial balance sheet of the institution. Under equitable insolvency payment of the defaulted item is sufficient to cure but under the balance sheet approach cure requires payment of a capital sum. It should also be realised that most companies cannot fully pledge their future cash-flows to raise current capital sums. In other words balance sheet insolvency approach is likely to overstate the level of equitable insolvency. This phenomenon is the subject of much recent and current academic study.
The question is which of these regimes is the more appropriate for the inter-temporal, inter-generational institutions that UK DB pension schemes constitute. It can be seen that schemes regularly continue to operate while in a state of deficit. In other words, they are reliant upon the existence, but not the exercise, of the sponsor balance of cost guarantee, the provision of future contributions by the sponsor. The relevant form of insolvency for a UK DB pension scheme is equitable insolvency, not balance sheet insolvency; clearly it was the view of the Finance Director interviewed.
Prior to sponsor insolvency, though the scheme only benefits from future contributions when they are made, it may operate soundly while in deficit. One of the features of scheme valuation under current accounting standards is that, for any particular deficit, the lower the discount rate used to present value liabilities, the longer a scheme may exist before it runs out of funds and defaults on pension payments.
Whether a scheme which is in deficit is insolvent prior to formal sponsor insolvency depends upon the answer to the question: can the sponsor be expected to make payments sufficient to repair this deficit prior to the sponsor’s insolvency? This is a standard guarantee evaluation. But notice that the scheme would still only be funded at 100% of its best estimate of liabilities and perhaps unable to run-off in orderly manner.
This raises the issue of when a guarantee can and should be called upon. With a standard conditional guarantee it is necessary for the trigger event to have occurred for any call under the guarantee to be enforceable. In this pensions case that would imply that the scheme should be unable to make the next payments to pensioners. The pension sponsor balance of cost guarantee may be called upon recurrently by the scheme, which eliminates some potential complications as to timing. It is notable that the Pensions Regulator bases its deficit repair calls on the sponsor based upon the premature balance sheet deficit view; this is costly. Indeed, European regulation, in setting a 100% funding objective at all times, is taking these calls to their extreme and most expensive limit.
Calls under the pension guarantee can only weaken the sponsor’s financial condition. This implies that trustees should evaluate the position of the scheme with and without the call. We shall revert later to this aspect.
While these balance sheet and equitable insolvencies are necessary conditions, they are not usually sufficient for a scheme to begin insolvency action against its sponsor. There may be further issues to be considered, such as whether the situation after recoveries from the sponsor’s insolvent estate is better than may be achieved by the scheme exercising forbearance. The concern here is that allowing the firm to continue may result in higher recoveries for the scheme than liquidation, with all of its attendant costs. In other words, it may even be in a scheme’s best interests to allow the sponsor to cure equitable insolvencies as they arise, rather than recapitalise the scheme. This is greatly complicated by the presence of the Pension Protection Fund.
The focus on the balance sheet view, and problems with it, can be expressed in terms of the Merton credit model. This is also intrinsically the model at the heart of solvency capital regimes. Under such a model, assets are modelled with drift and volatility, in more common-place terms, an expected return and some variability. The liabilities are estimated and the likelihood of assets being worth less than these liabilities at some future date is calculated. Additional current capital, that is further asset holdings, are required if this likelihood is higher than desired.
The central problem with this is that it does make any allowance for future actions. If it proves to be the case that developments are negative in this interim time, it is rational for the management to intervene by lowering the volatility of assets, recapitalising the firm or indeed managing liabilities in any ways which may be possible. A specific illustration is appropriate. In early September I visited Stockholm and spoke to a group of pension scheme managers. These were all members of the mutual pension indemnity insurance company PRI Pensiongaranti and insured by it. The policy covers scheme members against the insolvency of their sponsor employers. I explicitly raised the question as to what they would do if the situation arose that PRI Pensiongaranti sustained a large loss and was insolvent. The response was immediate and unanimous: “We would recapitalise it”, and continued: “You have to understand that this company is worth far more to us than the value of the pay-off under the insurance policy. It has considerable value to us prior to that happening.” Now under Swedish accounting practices, this pension indemnity policy is recorded in books and accounts at zero value, so this cannot be the motivation. The scheme manager was referring to the comfort that such insurance guarantees offer and the relief from pressure for action that it affords when times are difficult. In the UK such policies would also have an explicit value in scheme accounts, if the recommendations of the Insurance Working Party of the Pensions Reporting Accountants Group are followed. In more general terms and under most future states of the world, the sponsor does have an option to refinance the business. The Solvency II type of approach is incomplete and errs in a more costly direction.
Under the balance sheet approach, it is, of course, critical that both assets and liabilities are valued accurately. This introduces the current accounting standard. Assets are valued at market prices. By contrast, the liabilities are taken as the present value of projected future pension payment cash-flows. Trustees are encouraged to be ‘prudent’ in their choice of assumptions use to derive these projections. Guidance from the UK Pensions Regulator suggests that this ‘prudence’ term means that estimates should be conservatively biased relative to the best estimate. This means that the future cash flows are biased overestimates of the true liabilities ultimately payable. It seems that this bias is material – it is not uncommon for the projections to be overstatements of the order of 15% or more of the best estimate. This renders them of little value to the sponsor as this bias is time variant. It is particularly unfortunate that by operating upon individual assumptions it makes the whole process opaque. It would be superior simply to arrive at a best estimate and then add a margin to that – preferably after discounting to a present value. It makes estimation of the true effects of pension scheme risk exposures not comparable to estimates of the sponsor company’s own exposure to those risk factors. This joint exposure and sensitivity is critical for sound risk management of the sponsor company. In the UK it is this joint exposure which determines the insolvency likelihood of the sponsor and it is this insolvency likelihood which is the sole risk actually faced by scheme members, until the sponsor becomes insolvent. Let us not forget that protection of scheme members is the objective of regulation. The EC Consultation however is framed in terms of the scheme, not the sponsor.
Under the current accounting standards, the discount rate to be used in bringing the projected cash-flows to a present value under the accounting standard is a AA corporate bond rate or the equivalent gilt rate. This is singularly inappropriate and results in numerous paradoxes. It is perhaps simplest to explain the problem by analogy. Suppose that I wish to buy a house and need a mortgage to facilitate this. I will go along to the bank and the bank will typically offer mortgages at 80% of the market value of the house provided that I can convince them that their loan to me is secure. This loan is directly comparable to the promise of a pension to an employee; the question of one of security. The bank will tell me the terms on which they will make such loans, the interest rate charged and repayment schedules. They will then ask my earnings, current and prospective, since this will determine whether or not I can afford the loan and be expected to honour its terms in full and timely manner. They will not ask what my neighbour earns or what the market yield of gilts or corporate bonds is – they are irrelevant to the question of security of their advance. So it is also with companies, it is their earnings which determine the security of their long-term liabilities. The accounting standard is simply nonsense. It is interesting to note that the new Dutch regulatory regime allows schemes to use the expected return on assets as their discount function for liabilities. This is appropriate as most schemes there are industry associations and effectively stand-alone institutions lacking an explicit sponsor balance of cost underwriting guarantee. It is also notable that these schemes are effectively capitalised through the ability to remove indexation of member benefits – this makes them member mutual in nature. Similarly if the UK public sector ‘cap and share’ arrangements are ever introduced, this will make them member mutual in nature. The position in the UK where schemes do enjoy the sponsor employer’s balance of cost guarantee means that the correct discount rate for the valuation of projected pension payments is actually the employer sponsor’s long-term rate of earnings.
In order to justify regulation it is necessary first to demonstrate that there is harm under the status quo. In the UK, the long-term insolvency rate among active companies is 0.6% p.a. Its all-time high was 1.6% p.a. in 1991/1992. Since then (2002) the UK has introduced a new insolvency regime which appears to be marginally more debtor friendly. The PPF in its last annual report stated that its population of schemes covered had an average insolvency likelihood of 0.4% for the year 2011. The active company population experience for this year has so far been 0.7% only marginally above the long-term average. This lower than population average performance, for companies with DB pension schemes, is persistent. These companies are older, larger and more mature than the population average. It is newly founded companies which bias the statistics; fifty percent do not survive their fifth birthday and there is a high incidence of insolvency among these. In Europe, pension protection insurers such as the German PensionSicheringsVerein and the Swedish PRI Pensionsgaranti report similarly low rates of insolvency in their populations of companies covered – less than 0.5% p.a. It should not be forgotten that companies are actually twice as likely to be acquired, merge or wind-up solvently as they are to fail insolvent. The Solvency II regime sets a 0.5% p.a. likelihood of insurance company insolvency as its benchmark, but it is far from evident that this is materially better than the current and long-term experience of the UK or European corporate sectors.
In consequence of this, any application of Solvency II must be low cost to the sponsor if it is to avoid increasing the financial fragility and insolvency rate of the corporate sector. The capitalisation regime applicable to insurance companies would raise this corporate sector insolvency likelihood materially. This has harmful consequence for the social welfare more generally. In fact, in the UK, capitalisation at the level of the section 75 valuation is not feasible. This would require the UK corporate sector to make good deficits of the order of £650 billion to £850 billion, assuming an average current deficit of 20% of liabilities. The UK PNFC sector currently possesses £1.35 trillion of equity capital. UK PNFCs do currently hold £800 billion in cash, but if that were utilised, it would raise the issue of how any economic recovery or growth would be financed. The disruption in distribution that would arise would destroy the UK private sector since this would not be directly available to the majority of companies for refina