This website is entirely my own creation and has nothing to do with any former employers. Not only do I accept all responsibility but also I assert all rights over site content. Nothing on this website constitutes “advice”.

Although I am using this site as a portal for other sites, I have decided to include my “long term mantra” briefly. The longer version is at discount rates.

In the context of UK pension schemes (“final salary” or “career average”), I strongly believe that the “mark to market” funding approach is fatally flawed, leading to huge economic damage. The capital required has, unnecessarily, weakened many firms of all sizes and too many people have lost their pension expectations. My primary emphasis had been on final salary pension schemes valuations. More recently, however, I have realised that the true problem is as basic as the discount rate. Quite simply, discounting future cashflows over very long periods is pointless if we can't adequately estimate what rate(s) to use - which we can't.

If anything, actuaries claim to understand risk, with an added sub-text that being “broadly right” is better than being “precisely wrong”. Very long-term financial entities, such as DB pension schemes, need not be restrained in the same way as short-term financial entities, because assets can't simultaneously be aimed both “long” and “short” (without a premium). Discounting future cashflows is not always that helpful because risk quantification is very poorly captured by scalars, which are all discounting can deliver. Instead, stochastic processes are a much better approach, for which there is very much greater capacity than in earlier years. In order to be prudent, one needs to be aware of the best estimate, a basic economic concept not apparently recognised within the pensions industry. Paraphrasing from Redington (1952), “derisking” leads to “derewarding”, which is economically wasteful.

Financial markets demonstrate at least two specific weaknesses, namely being prone to herding (following trends for too long) and an inability to price tail risk (or even to perceive it). Over the long term, therefore, it should not be surprising that using mark-to-market tends to be far less efficient than going off-market (with more relevant data captured). Further, market prices have no predictive return power (Fama, 1965).

Single numbers (scalars) are not appropriate for representing many future uncertainties. That is especially the case when we don't even say what the scalar signifies (mean? median? mode? specified percentile?). Instead, we should be looking at multi-dimensional results, preferably with confidence intervals, which cannot be done with deterministic approaches.

Using discount rates is actually nothing like as helpful as carrying out robustly backed stochastic projections, needed in order to be able to advise the sponsor and the trustees properly (DB pension space and others). Capitalisation and the associated discount rates are neither necessary nor helpful for long-term projects.

Actuaries (financial scientists) need to follow evidence, interpreting evidence as experts. Rather than being objective, interpretation is subjective. There is nothing wrong with subjectivity, so long as independence is demonstrated (not merely rooted in groupthink); the evidence available is taken into account; and full, cogent explanations are provided.

If we can't get relatively simple stuff right, how can we suppose we can do more complex stuff? The “best” really can be the enemy of “good” (Confucius).


Jon     06 December 2018