A simple question of defining an oft-used term leads to important but tricky questions, and throws light on key issues. It becomes clear that running a pension scheme in run-off mode involves a very different set of considerations to deficit recovery mode. Mindset, language and tools need to change. The conclusion is that a full definition of self-sufficiency needs to capture both funding and investment policies (either defined separately or jointly), and make allowance for demographic risks and all expenses.
As someone once said in a different context: “it’s hard to define, but I know it when I see it”. The speaker concerned may well have been talking about the concept of self-sufficiency as it applies to DB pension funds, although I’m not sure they were.
With c50% of DB schemes according to data from both Aon and Mercer (see left), targeting self-sufficiency as their long term objective, it seems important to make progress on some sort of definition, or framework for one, and today there is relatively little readily available thinking out there on the topic. The 2018 DB White Paper “Protecting Defined Benefit Pension Schemes” didn’t mention the term once in 76 pages (buy-out was mentioned 16 times and consolidation 56 times).
Historically, the old actuarial paradigm was to discount pensioner cashflows at gilts +0.5% (a relatively small margin), thereby envisioning a future all-pensioner world where the funding basis would permit investment in pretty low risk assets. Job done! While this paradigm has certainly caused a lot of anchoring (see for example the common discount rates used in the Mercer 2018 European Asset Allocation Survey on the right), and is not entirely wrong in many ways, it certainly seems like it is due at the very least for a refresh. We need a new paradigm.
What is self-sufficiency?
You could argue, and many have, that self-sufficiency is either a spurious concept, or over-used and unhelpful term (or both). This may be, but with 50% of schemes saying they are targeting it I think we are stuck with the terminology for now and may as well figure out how to interpret it, rather than try and invent new terminology.
The Pensions Regulator in its 2018 guidance makes the following comments on self-sufficiency (emphasis mine):
When a pension scheme reaches a certain level of assets (the self-sufficiency level), it expects to be able to sustain itself by investing those assets on a low risk basis and pay members’ benefits as they arise without any additional support from the employer. Very large pension schemes may view this as a more cost-effective alternative to a buy-out with an insurance company.
Under a self-sufficiency approach, the ongoing reliance on the sponsoring employer is kept at a minimal level. This requires a low-risk investment strategy to minimise the chances of the employer having to make good any investment losses.
Typically, the investment approach may be similar to that of an insurance company under the buy-out approach, but without the solvency requirements of an insurance company and the need to make a profit for shareholders. Consequently, the liabilities under a self-sufficiency valuation should be lower valuation.
At the heart of the concept of self-sufficiency as it is commonly used I believe is something like:
An ongoing state where the scheme continues to exist and pay pensions, with a reasonable expectation of no further financial contribution or dependence on the sponsoring employer.
Note this doesn’t necessarily mean that the sponsor no-long exists, just that the scheme can effectively exist independently from a financial perspective, it doesn’t depend on the sponsor for further contributions.
This definition implies a few things:
- The scheme would need to be well funded, on a conservative basis incorporating a buffer over best-estimate parameters (so as to have a reasonable expectation of requiring no further contributions, even in times of adverse investment experience )
- The scheme would need to be invested in a very low risk way, in order to not be exposed to significant falls in the funded position, which would create a need for more contributions (however I believe there is a general understanding that self-sufficiency is not “no risk”)
- The scheme would need to have margins built into the demographic assumptions to allow it to withstand some adverse experience (if these are not explicitly hedged)
- The funding assumptions would need to incorporate future expenses and PPF levy payments
So it is clear that both funding, and investment policies are key ingredients of how to define self-sufficiency, rather than it being defined purely in terms of funding policy. Secondly, handling demographic risks will be a key component, and building in enough margins for adverse experience to be truly self-sufficient. Finally, ensuring that future expenses are fully built in is also key.
Extending that line of thought, we could choose to define self-sufficiency explicitly in terms of particular funding and investment parameters (eg a discounting basis and investment risk level), or we could choose to define in terms of a more general parameter that is itself determined by the funding and investment policies. One potential option here could be a modelled metric such as the Probability of Paying Pensions (for more background on such measures read Redington’s research piece Meeting Member Promises).
This has the advantage of somewhat elegantly capturing both funding and investment policies into a single number, but a few disadvantages in terms of being unavoidably model dependent, with different models yielding potentially quite different results, and potentially introducing model risk into the equation, and even being “too” precise with trustees perhaps feeling under pressure in the choice of one level 95% say, instead of another higher level.
The table below illustrates the Probability of Paying Pensions (coloured cells) for a range of funding assumptions, assuming no further support from corporate sponsor. Source: Redington Integrated Risk Model
Notes: investment strategy targets returns 0.25% above funding rate, strategies vary. Longevity risk not included (model can be extended to include longevity risk).
The PPF adopt such a probabilistic approach in calculating the Fund’s “Probability of Success” in their long-term funding strategy, in the 2018 update the PPF assessed a 91% probability of the fund being able to meet all future payments (an increase from c80% in 2010). In its funding strategy the Fund estimates that a 10% “self-sufficiency margin” over the best estimate funding assumptions is required to achieve a 90% Probability of Success. The PPF’s section 179 valuation basis discount rate is effectively gilts+0.2% pre-retirement and gilts-0.2% post retirement. The PPF notes that to give a 99.5% confidence of paying all benefits a margin of 50% would be required. These are useful reference points (although the PPF is in a different position to all other DB schemes, being exposed to the risk of additional claims on the PPF in addition to the usual investment and demographic risks, as well as having the ability to raise capital through the levy).
Defining the funding basis for self-sufficiency at a certain level, perhaps with reference to a best-estimate plus buffer approach feels pretty familiar given that is what we are used to from the “old” paradigm, but an interesting question becomes how prescriptive you want to be on the investment side. What is the long-term goal on the investment side, and should this be built in explicitly into the definition of self sufficiency?
Many self-sufficiency conversations tend to imply that the goal is investment in a portfolio of high-quality corporate bonds, probably something with a cashflow match (whether such a thing can be found, created, or synthesized) in order to defease (or directly offset) the liability cashflows. Indeed that is similar to what TPR is getting at with the statement “Typically, the investment approach may be similar to that of an insurance company under the buy-out approach, but without the solvency requirements of an insurance company …” And indeed there doesn’t tend to be much objection to investing in a portfolio of investment grade corporate bonds, these ought to reliably provide the cashflows needed in almost all circumstances. This sort of investment approach will also minimise cashflow risk and re-investment risk (two types of risk that tend not to feature too highly when considering a DB funding in deficit-recovery mode, but do play a larger role when it comes to self-sufficiency).
But cashflow matching might not always be the right choice. Despite being conceptually simple it is quite complex to implement in practice, potentially involving considerable use of derivatives to inflation-link all cashflows and translate foreign currency cashflows into sterling. Suitability of a cashflow-driven approach may well depend on a scheme’s size, maturity and the level of credit spread. Credit spreads do vary quite a lot through time and we ought to consider scenarios where the credit spread becomes very low indeed, would we consider investment portfolios that didn’t focus on credit, perhaps a “barbelled” portfolio of majority gilts with a small slice of return generating assets such as equity or illiquids? what about a portfolio of bonds but not one that focused on matching each cashflow exactly?
As soon as we move away from a cashflow match there are a lot of possibilities, these will also introduce potential reinvestment and cashflow risks, which need to be weighed up. Should our definition of self-sufficiency leave room for these sort of trade-offs to be evaluated? Perhaps we ought not to be too prescriptive on the investment side after all – it could end up causing concentration of risk in particular asset classes (for example, corporate bonds). Perhaps the investment policy could be “handled” at a higher level by prescribing a level of risk, either quantitatively or qualitatively.
Unfortunately, both quantitative and qualitative definitions both have drawbacks. A quantitative definition can seem over precise and very model dependent (” A funding ratio at risk of 5.5% at the 1 year, 95% level”), and any metric that didn’t include longevity risk would risk being effectively meaningless in the endgame when longevity is likely to be the largest risk. Indeed, you could even argue that with longevity risk present, there is no need to try and get the investment risk down beyond a certain point, or be too precise about it, as there are likely to be theoretical diversification benefits between investment and longevity risks. On the other hand a simply qualitative definition of risk ( eg “low risk / very low risk / no market risk”) might feel nice but is always going to be pretty intangible and leave a wide variety of interpretations on the table – which might be a good thing.
A final question (for now) is that regarding risk to the demographic assumptions such as longevity. Self-sufficiency is framed around paying pensions, but investment and funding tend to be framed around a particular projection of the cashflows at a certain point in time. These two are not the same, and the difference is really important when it comes to the endgame as it could pose the largest risk to paying pensions (for example, see the risk impact illustrations below). How to handle this difference?
The chart below illustrates a risk breakdown for a highly diversified portfolio aiming for a relatively low level of excess returns. Longevity is the largest single risk factor
To be truly self-sufficient we need to make some sort of allowance for longevity and other demographic risks – we can’t base the definition of self-sufficiency simply around meeting the projected cashflows as they stand today. So some reference to longevity hedging or reserving policy is probably required to complete the definition. Again, this could be quantitative and model-dependent or qualitative and vague. Or it could be comprehensive and state a full longevity hedge as part of the self-sufficiency definition. An alternate approach could be “reserving” – in effect preserving a buffer in both funding policy and investment returns to allow for future adverse changes in longevity. Again this seems quite key to the core definition of self-sufficiency.
In this piece I set out to try and make progress on a definition for self-sufficiency, it is clear when thinking through that the challenges facing a self-sufficient DB pension fund are quite different to those facing one looking to close a funding gap, things like re-investment risk, cashflow risk and longevity risk loom much larger and warrant specific measures to address them.
This article ends with more questions than it answered, but hopefully it has been helpful in illuminating some of the knotty issues in actually pinning down a definition for self-sufficiency which is something that at least 50% of DB schemes are going to need to do. To some extent this endeavour is philosophical in nature – similar to investment beliefs and it probably depends on and flows on from the investment beliefs of a particular group.
Conclusion: A full definition of self-sufficiency needs to capture both funding and investment policies (either defined separately or jointly), and make allowance for demographic considerations either by reference to hedging or reserving.
For more information on DB endgame planning see Redington’s ampersand institute recent paper, Destination: endgame.
Dan is a member of the self-sufficiency, buy-out, and consolidation working party of the Institute and Faculty of Actuaries. One of the aims of this group is to deliver a practical reference document for actuaries on the issues and frameworks they need to think about when advising clients on the end game.