Quest for the ‘golden’ discount rate
An outsider considering the discussion on the appropriate discount rate for discounting future pension payments would be amazed at the large number of those involved in the debate as well as at its length, depth and breadth. To mention but a few of those involved: the European Commission, the European Insurance and Occupational Pensions Authority (EIOPA), local regulators, economists and actuaries and the International Accounting Standards Board (IASB) and its Interpretations Committee, have been ‘searching’ for the treasure of this ‘golden’ discount rate for years. It feels like the search has partly been conducted in the limelight and partly in the dark. In any case, many attempts have been made at approaching the quest using ever more inventive and sophisticated arguments. Without much success, it may be added.
In its simplest form, the discount rate is the rate used to convert an amount of money payable in the future into a present value. In other words, the discount rate converts the value of an economic asset or liability expressed in ‘tomorrow’s currency units’ into an economic good expressed in ‘today’s currency units’.
In the areas of pensions, life insurance and other liability and asset valuations the use of a discount rate is a powerful tool to convert not only a single amount due in future but, rather more typically, a complex array of amounts payable in future into a single present value.
It should be clear that an amount that is payable in one or two weeks from now should have a similar value today as it has ‘tomorrow’. If that timespan is extended significantly, however, the effect of the discount rate on the present value may be potent. For example, an inflation-linked deferred pension for a 50-year-old pension plan beneficiary will typically have an average weighted payment date of some 20 years or so. Using a discount rate of 5.5% as a reference amount, doing the calculation at, alternatively, 3.5% will result in an increase of the liability of some 50% vis-à-vis the reference amount. So the discount rate used can affect the resulting present value very significantly.
There are two areas on which the discussion is currently focused – accounting and regulation.
In its fundamental overhaul of IAS19 in 1998, the IASB prescribed that, for accounting purposes, the selection of the discount rate for discounting future pension obligations was to follow the return on “high quality corporate bonds” of the same currency and duration as the obligations, provided that there was a deep market in such securities. Last year, the Interpretations Committee of the IASB was approached by the German accounting standard setter to provide guidance on the interpretation of “high quality”. During its February meeting this year, the IASB confirmed current general practice of looking to AA-rated corporate bonds to be high-quality. However, the decision may be regarded as guidance only on the current standard, subject to its fundamental review in the (more distant) years ahead.
Apart from this stop-gap decision for pensions, the IASB decided in December 2012 to take on its three-year agenda a research project on discount rates aiming at finding common principles for financial statement purposes that could be applied for discounting all sorts of different future cash flows, not only those relating to insurance and pensions.
On the regulatory side, the European Commission has extended its scope from insurance to also encompass European pensions. Since its first proposals aired for insurance in 2004, the Commission has put forward many different approaches, some seemingly quite arbitrary.
Why is it so difficult to define a discount rate that is universally applicable? The unsurprising reply is that the answer depends on what objective the calculation is to fulfil, which in turn depends on the perspective of the party that intends to put the calculation to use. In addition, there are large sums of money involved, all dependent on the discount rate used. There are many different schools of thought as to what approach should be used to fulfil any given objective. Oversimplifying a little, I believe the main approaches to be:
• Expected return approach. When considering the solvency of a pension fund, some argue that it is intuitive to discount the liabilities at the same rate at which the plan assets are expected to earn returns in future. This is because future payouts are to be met by available assets.
One advantage of this is that it takes account of the underlying investments that the pension fund is actually holding. A disadvantage of the approach is that it is difficult to choose an expected return that can be accepted by outsiders as being objective and consistent with the view taken by the financial market(s).
In the past, this approach has often failed – mostly due to the anticipation of excessively high returns on plan assets – and had then to be appropriately adjusted, sometimes painfully, as was the case in the UK before the inauguration of the Pension Protection Fund in 2004.
• Debt instruments approach. The reasoning is that future benefit payments can be compared with future payments from debt instruments. Internationally, this approach is currently the most popular in accounting and regulation. It assumes the parties to the assets and obligations being discounted, are risk neutral. In general, market yields on fixed interest bonds or swaps are readily available and can be used as references
Although, based on this same starting point standard setters and regulators have, however, chosen either a point-in-time approach or have determined that long-term averages should apply. On occasion, the market yield is further adjusted to implicitly allow for prudence.
One advantage of this approach is that the rates are fairly easily determined from market prices, thus enabling a high degree of objectivity. One disadvantage is the subjectivity involved in setting the appropriate reference credit rating of the debt instruments to be used, how these are to be interpreted, the periods over which averages are taken as well as any applicable adjustments. For example, in setting the appropriate level of credit rating, is it clear to all concerned that the application of a risk-free discount rate implies that a particular corporate pension promise must be less subject to default than an AAA-sovereign creditor?
In the recent past, this has resulted in increasingly low discount rates in many currency areas, mostly due to quantitative easing policies adopted by those countries’ central banks. In response some simply adjusted upwards the rates derived from the debt instruments approach used before the change. Recent examples include the regulators in Denmark, the Netherlands and the US.
A further theoretical requirement for both approaches to work is that the nature of both liability and asset cash flow streams should bear substantially the same characteristics, in particular the risk of actually being paid. Even if this requirement seems obvious, it brings with it thorny practical questions.
Market-consistent, market-related or plainly subjective?
The main justification for choosing rates that are market-consistent is the objectivity of prices in a sufficiently functioning market. However, events in the last 20 years have shown that this underlying assumption may at times simply be wishful thinking. Examples include the post-2007 credit squeeze and the recently-discovered LIBOR fixing scandal.
A market-related approach appears to be objective but isn’t. Since even a little arbitrariness is still arbitrary and can lead to a lot of effort being expended to achieve spurious accuracy. This is where I personally believe the Commission has been heading with its various ideas for regulating European pensions.
Subjective approaches can nevertheless be useful and well-balanced, if one accepts that an element of arbitrariness is unavoidable – it is better to be approximately right than precisely wrong. With regard to the European pensions regulatory debate I would propose that allowing national regulators to continue to exercise their judgement differently from one another is probably the best route to take. This should avoid the dangers of herding and would allow the different pensions markets to continue to evolve naturally.
Is there a ‘golden’ discount rate?
I fear that the quest for this treasure will never conclude to the satisfaction of all. One reason for this may be that the answer probably doesn’t exist or doesn’t matter as much as some believe. Whatever route is chosen by a standard setter or regulator, I believe that the following factors are significant:
• The reasons for choosing a specific approach for discount rate selection should be explicitly disclosed in detail by the regulator, or standard setter.
• Harmonising the discount rate across Europe for regulatory purposes of insurers and pensions will have a significant effect on financial markets; treating pension funds and insurers identically in this respect will unjustifiably amplify the effect.
• The discount rate should be chosen to take account of the underlying business model of the entity making the promise as well as the legal nature and other characteristics of the benefit itself. This is not a trivial exercise within a country let alone across Europe and the world.
• Before determining that a risk-free discount rate should be used, regulators should carefully consider whether requiring corporates and their pension funds to pretend that they are better guarantors of their pension promises than their respective sovereigns, even if they are AAA-rated, really makes practical sense.
Based on the futility of finding the golden discount rate that will answer all problems for all parties, we may ask the question whether its selection actually matters that much. Is it not possible with today’s computing power to move beyond the arcane present value calculations towards assessing solvency on the basis of projected asset and liability cash flows and comparing their relevant characteristics in doing so? If the requirement for producing present values cannot be jettisoned yet, perhaps a suitably detailed cash flow analysis can be used in parallel. Most importantly, it may be a case of actually accepting that we cannot foresee the long term future, however much we would like to.
Alf Gohdes is head of actuarial consulting, Towers Watson Germany