The £1bn myth: how scheme-level costs change with size

Building scale in pensions is a commonly cited goal by policymakers. The logic that bigger schemes incur lower costs makes logical sense, but is not often tested in practice.

In an attempt to do so, we need to bring together the conclusions of the articles in this series so far, which I have summarised as:

• Different asset classes have different proportions of explicit to implicit costs, and only some asset classes attract performance fees.
• Asset allocation changes as schemes increase in size, with an apparent discontinuity occurring at about £1bn in assets. Schemes below this point rely heavily on passive equity, while larger schemes above use more fixed income and liability-driven investments, as well as relatively high allocations to illiquid and alternative asset classes.
• The use of segregated mandates increases as scheme size increases, again with the most dramatic increase in segregated mandates occurring within funds of more than £1bn in assets under management. Furthermore, the size of mandates increases as scheme size increases, and the total number of products used by schemes tails off as size increases.
• Most asset classes show scale economies at larger sizes of mandate, for both pooled and segregated mandates.
• For almost all asset classes, paying more does not generate higher returns. The exceptions are, in the main, asset classes that lean towards performance fees and are illiquid.
• Segregated mandates generally incur higher costs, but also generate superior performance (see table below).

Table 1: Comparison of total cost and net performance between pooled and segregated mandates of the same asset classes.

Source: ClearGlass Analytics

Where am I going with this? Well, if mandate size rises as scheme size rises, and the larger the mandate the lower the cost, this means that larger schemes should have lower costs.

Additionally, the shift from pooled to segregated mandates, and the change in asset allocation for funds of more than £1bn AUM, should result in a spike in costs at this level of assets, which should then fall again as scheme size increases.

At the same time, the shift to using segregated mandates by schemes of more than £1bn, and their higher allocations to asset classes that generate performance – and performance fees – should cause scheme-level performance to rise.

Finally, the proportion of costs that are implicit – compared with explicit – should rise in schemes larger than £1bn.

We have the data to verify this, and what I see is actually pretty close.

Larger funds tend to have lower annual management charges and lower total costs, but with the discontinuity occurring at £1bn and above as discussed.

Larger schemes also tend to attract more performance and property costs, in line with their increasing asset allocations to the asset classes that carry these cost types. The proportion of implicit costs tends to increase as scheme size increases, and net performance rises as scheme size rises.

Interestingly, at the scheme level the proportion of costs that are transaction costs does not exceed 33 per cent.

Table 2: Summary of mean costs by scheme size

Source: ClearGlass Analytics

Some disclaimers

The variability in the data is high, and what you are seeing here are averages. To pronounce that the total costs of pension management are X or Y, even for the size bands I have given, would be both irresponsible and wrong.

I am also certain that the size bands could be narrower, but even with more than 200 schemes in the dataset, I do not feel comfortable breaking the bands down further – there are approximately 5,400 DB schemes in the UK.

Moreover, the decision to shift a scheme from a mainly passive equity to a mainly fixed income asset allocation will be unique for each scheme and sponsor, and will probably have little to do with the size of the scheme in question and whether it has surpassed £1bn.

The good news is that we are currently collecting data for almost 800 DB schemes (perhaps more) with 10,000 or so mandates included in the 2019-20 cycle, and when we are done I will rerun the analysis to see what has changed, if anything.

I will also introduce real academic rigour into the analysis through the research group I have brought together under professors Iain Clacher, at Leeds University, and Tim Jenkinson, at Oxford. Who knows what their analysis will reveal, but we will all await the outcome with great interest.

Dr Chris Sier is chair of ClearGlass Analytics and led the FCA’s Institutional Disclosure Working Group