OK, so what is it? Does it even exist? Well, it's a bit like an elephant, which you can recognise when you see all of it. The essential criterion is whether or not, normally the latter, future cashflows must be secured in the near future with some other entity, with assets needing to be realised. In the past, for most pension schemes with contributions continuing to be made, cash flows would have been positive for some time to come. So long as the trustees and sponsor are advised that this should be the case for at least say 10 years, then I reckon that gives long enough for the trustees (and sponsor) to weather equity risks, with returns likely to be higher than on bonds. Sadly, so many sponsors have been scared into closing pension schemes that cashflows are even less likely to be positive. TPR’s 2020 DB Funding consultation and what is being introduced (subject to Mansion House) will definitely make matters worse.
Given the existence of predators (private or public), how confident can the trustees and sponsor be about how long they will have? So long as the issue has been discussed, the trustees and sponsor are entitled to form their own views and act accordingly. In particular, investing in bonds is unlikely to prove a panacea. It is far from clear that this issue is ever specifically covered by either the Statement Of Funding Principles or the Statement Of Investment Principles but I suggest it ought to be.
It is sometimes argued that it is necessary to focus upon market values in order to target discontinuance funding. However, if there is a long term, then the target is a needless boundary constraint. In reality, the argument is circular, with wide solvency swings, essentially due to mark to market, being taken as a reason for as speedy an exit as possible, implying discontinuance. That circularity needs to be removed.
Where the responsible stakeholders agree that it is reasonable to assume that they have the luxury of a long-term, say 15 years or longer, then equities are extremely likely to provide higher returns than bonds. This can be seen from history and likelihood.