Pension contracts and regulation
Robert C Merton and Jan Snippe argue that Dutch pension legislation should be inspired by fresh and logical thinking
The FTK (the current set of prudential regulations for defined benefit pension plans in the Netherlands) has brought many good things. Even so, it is now up for a revision. The ministry of social affairs and employment (SZW) and the central bank (DNB) are preparing this, and it is generally expected that, barring possible complications as a result of the recent cabinet crisis, the ministry will present its proposals in the coming months.
We can only guess what these proposals will look like but, based on recent statements by representatives of both SZW and DNB as well the recent recommendations of the Frijns en Goudswaard committees, it seems likely that the current imbalance between the required certainty of nominal and real pension rights will be rectified and that buffer requirements will be increased to reflect revised risk estimates. It is also possible that the importance of the so-called ‘continuity analysis' will be increased, to the detriment of the current system of solvency requirements.
Although such changes may well improve the supervisory regime, we suspect that a more fundamental revision is needed, especially now that many pension funds are in a recovery process and, as both Frijns and Goudswaard have recommended, pension funds may need to revise their pension contracts to reflect the fact that their members are more exposed to risk than the current contracts may be taken to suggest.
In our view, a more fundamental review should not be undertaken independently of the revision of pension contracts. As we will show in this brief appraisal of the current regime, pension products and regulatory requirements are closely related. The fact that current products have initially raised expectations that now prove unrealistic cannot be understood independently of the current supervisory regime. We conclude with a plea for a new type of pension arrangement.
New pension products needed
The need for a revision of both the existing pension contracts and the supervisory regime is a logical consequence of the current situation, with under-funding and insufficient reserves all over the place. The current distinction between so-called unconditional and conditional pension rights does not mean very much in this circumstance, and it has probably never been as meaningful as it was initially meant to be, the law (the Pension Act 2007) assumed it to be, and the existing contracts seem to suggest.
The existing regulatory regime cannot deal with the current situation. This is evident from the requirements that pension funds with insufficient reserves are supposed to comply with. They are required to recover their reserves over a period of 15 years. During that period, unconditional pension rights are sometimes significantly less certain than required (97.5% on a one-year horizon). Even so, pension funds are not required to revise their unconditional promises. Even pension funds with a solvency ratio of less than 105% have several (three, and temporarily, in view of the impact of the credit crisis, five) years to recover to that minimum ratio. Under such circumstances, unconditional pension promises have become more or less meaningless.
It gets worse if recovery plans assume asset returns in excess of risk-free interest rates without any correction for the risk that those assumed returns may not be realised. The law allows such plans, and DNB may approve them. Their approval suggests a degree of confidence (‘we are insufficiently funded, but we will recover, so that there is no need to worry') that does not seem to be justified. This is bad not only for plan members, but also for the supervisor because unmet expectations may ultimately turn against it.
Spreading risks is not sufficient
We also wonder why unconditional promises are associated with a 97.5% certainty level, rather than the 99% requirement that insurance companies are supposed to comply with. Not only because 99% is closer to the 100% level that really unconditional promises would seem to require but also because the 97.5% seems to assume that unconditional promises may be safeguarded through diversification rather than hedging or insurance. Developments in financial markets over the past few years have clearly indicated that risk spreading does not guarantee that losses will be avoided that render unconditional promises unsustainable. You may therefore have to cut it (through hedging and/or insurance) to ensure that unconditional promises can ultimately be met.
No such thing as a free lunch
It is likely that many readers will object to our seeming recommendation of higher probability standards than pension funds and their members and sponsors can afford. Similar objections were raised when the first proposals were made that ultimately led to the current regulatory regime. The proposals were for a 99.0% certainty level and required buffers to protect conditional rights (such as indexation). We appreciate such objections, but we have less understanding for the way they have been (and are still) dealt with. There are no ‘free lunches'. If you cannot afford a particular product, you have to select another one. People who do not do that, or who suggest that they may systematically offer products for less than their cost, are just fooling themselves and others.
Many pension funds have experienced such a dramatic weakening of their funding positions that their members are now exposed to materially higher risks than they, and the funds themselves, had anticipated. This is directly attributable to decisions taken by these pension funds. They have taken more risk than they could afford and charged premiums that were too low. The disappointing degree of certainty they now offer is therefore not a surprise. We have actually known about it for quite some time. Again, this is primarily due to choices made by the pension funds. For years, they have offered different pension products to those they suggested and assumed, and their members expected and perceived. This would not have been possible, however, if the law had not allowed it and the supervisor had sensed the risk at an earlier stage. A flawed (because inconsistent) supervisory regime allowed flawed pension products, which has ultimately caused a situation that, as Frijns and Goudswaard have indicated, cannot be sustained.
Promises without funding
The new pension contracts that Frijns and Goudswaard have recommended offer less certainty than the old contracts seemed to promise but failed to deliver. Even so, they retain a defined benefit character. This only makes sense if these new contracts clearly specify the probability that the promised benefits will eventually be paid and, depending on the contract, indexed for inflation. The regulator will have to ensure that funding and investments match that probability and indicate how much funding will actually be needed. The revised supervisory regime should plan for that.
Simply lowering the unconditional promises and simultaneously increasing the conditional rights will not be enough to solve the current problems. The flaws of the current regime would simply continue, even though it would be less restrictive. For a real solution something different is needed. The current dichotomy between conditional and unconditional liabilities will have to be replaced by a broader spectrum of possibilities, each with a different degree of certainty and corresponding buffer requirements. The result may be similar to what the initial proposals that ultimately led to the current regulatory regime seemed to imply by requiring buffers not only for unconditional liabilities but also for conditional liabilities. Unfortunately, these proposals did not indicate how to implement the idea because the regulator admitted that it did not know how it could be done. It seems unlikely that the situation would be any different this time. Promises about the certainty with which benefits may be realised do not make a lot of sense if they are not combined with clear funding requirements to match them.
Could this be solved by changing the current emphasis on solvency requirements to focus on longer-term continuity tests? If such tests are based on stochastic models, they would at least allow us to judge whether the probability of any promised benefits is in line with the initially agreed level of certainty. If the probability turns out to be lower, however, we will end up in the same situation we are currently experiencing with respect to pension funds with insufficient funding or reserves below the applicable buffer requirements. In other words, putting more emphasis on continuity tests does not provide a real solution. DC is inevitable.
This leaves us with no alternative other than to accept that there is no point in promising a fixed probability of realising certain benefits. With this acceptance, however, we basically end up with defined contribution (DC) arrangements. DC plans do not require prudential standards. Instead, they require pension funds (and the regulator) to determine whether risks are taken that do not correspond with the risks that participants knowingly tolerate.
In addition, the regulator may be asked to determine whether any statements made about probabilities are adequately supported by the available evidence. These are, at least to some extent, the kind of concerns that are implied in the so-called duty of care principles.
Unfortunately, these principles are based on assumptions that do not seem sensible. First, they assume that well-informed people make good decisions. Regulators across the planet are increasingly inclined to increase their requirements as to the quantity and quality of the information that pension plans provide to their participants. This assumes that participants are capable of responding rationally to the increased transparency, and to perform analyses and make decisions that they did not have to make in the past, are not capable of making now, and are not likely to execute effectively in the future, even with attempts at education. Furthermore, in the absence of the necessary products (for instance to protect against possible lost indexation of pensions), participants will not be able to take risk-reducing actions even if they understand the complex conditional benefits of the new contracts.
The current duty of care principles also seem to assume that investment risks are lower for longer time horizons. This assumption is based on the observation that longer-term volatilities of returns tend to be lower than short-term volatilities. This implies that the probability of shortfalls for longer periods will be lower than for shorter periods. It is wrong, however, to use this as a risk measure, because it ignores the size of possible shortfalls and therefore also ignores the fact that this size increases with longer time horizons.
As a consequence, current regulations and supervisory practices allow products that require a significantly larger involvement than participants can deal with and offer investment solutions that are entirely inadequate. This makes them not only unnecessarily risky for the participants, but for their employers and plan providers as well.
Unnecessary risk can be avoided, as is evident from a new plan design called structured DC that provides participants with choices that allow them to arrive at a sound retirement plan without becoming financial experts. Structured DC solutions propose a clear target level of pension income that is protected against interest rate, inflation and longevity risks, subject to a minimum level of income to reduce the risk of ending up with an unacceptably low pension. Each participant has an investment account with investments allocated in a manner designed to achieve either a default or an individually selected post-retirement income target. The management of the account is arranged collectively and does not require the participant's active involvement. Participants are, however, permitted to personalise their targets and adjust their contributions and retirement date. The approach has been put into practice in Europe and has performed well even under difficult market conditions.
If structured DC were adopted, a possible responsibility for the regulator would be to determine that contributions are invested as suggested; that is to maximise the probability of realising the desired income subject to minimisation of the risks of ending up below the selected minimum income. In addition, the supervisor may monitor whether providers check that plan participants who opt out of the default settings consider a minimum income level that is lower than the default to be sufficient for realising an acceptable standard of living in retirement. In addition, plan providers may be required to ensure that participants are alerted if the probability of realising their preferred standard of living falls significantly below their initially selected probability, so that actions can be taken to remedy the problem.
Structured DC plans are compliant with current regulations in the Netherlands and they are also compliant with the more recent interpretations of those regulations by the regulator (the AFM), although those regulations are not yet required. Obviously, it is not up to us to suggest that they should be required. However, when regulations continue to allow flawed plan designs, the risk will remain that participants are exposed to bigger risks than they can afford, and that the dramatic losses that many of them have experienced over the past few years will re-occur. This could be avoided with better plan designs. If the government wants to achieve real improvements, significant changes in legislation will be necessary. Legislation should no longer be based on outdated and flawed economic theories, but rather be inspired by fresh and logical thinking - the same sort of thinking that has led to the development of structured DC.
Robert C Merton and Jan Snippe are associated with Dimensional Fund Advisors.
Robert C Merton is also John and Natty McArthur University Professor at the Harvard Business School