top of page
Search
  • William Bourne

LGPS Cost Data – Analysis by Pool



My article questioning the cost savings which the Government is claiming from LGPS pooling received more hits on LinkedIn than I have ever had before, except when I nearly killed myself in a car accident.


Readers may well be interested in the analysis by pool. I used the England and Wales SF3 data collected by DLUHC and categorised each fund by the pool they have joined. I then calculated the investment costs as a bps of AUM at the end of the period. Here is the data. The two right-hand columns show the five-year annualised increases both in absolute terms and in bps terms (i.e., allowing for the increase in assets):


Let me start by saying that the data needs to be treated with care for many reasons:

  • The costs of pooling may have been allocated to investments or to governance.

  • Investment costs may only include directly charged investment fees or include (as is now guidance) an estimate of fees which are embedded in an investment product.

  • Different pools and funds have different investment models. Some may have much more in ultra cheap passive funds, others in expensive alternatives.

As a check on the first two of these I made the same calculation on total costs and found a similar story.


There are also many reasons behind the doubling of investment costs over this five-year period, some to do with pooling, others not.

  • There has been a clear trend to more expensive private market alternatives such as infrastructure and private equities.

  • There has been a trend towards ESG-friendly investments, which again tend to be more expensive.

  • The set-up of pools and the increasing demands on LGPS pension teams has required greater resources.

  • Private equity in particular paid out bumper performance fees to General Partners in 2021 and 2022.

  • The scale of pool partners ranges vastly from an average AUM of £19.4bn (Northern) to £1.5bn (London CIV).

  • Funds may have during this period started to include investment fees previously embedded in underlying products.

That all said, I can draw some conclusions from this data.


A good international target would be for costs to be below 40bps. This is in line with a number of studies over the years, including one commissioned by DLUHC’s predecessor in 2013* and a broader survey of 330 international investors. Only the Northern pool comes in below that bar, partly because of the size of its partner funds. Scale does help, but note that the next lowest cost is the smallest pool, Wales PP.


The model of renting an ACS structure used by Wales PP and ACCESS seems to have delivered lower costs, although that disregards the risks around LINK’s resilience and may not be true in the longer-term. Outsourcing investment expertise to a private consultant, as Wales does, can also keep costs down.


LPPI’s costs are more than double those of the other pools and have grown faster than all except for Northern. Some of that will be due to its London location and its asset allocation mix (and the fact that it is closer to a fiduciary model than other pools). But it seems to be a case where greater scale is needed.


The elephant in the room is that cost savings are dwarfed by the impact of performance. All pools need to ensure they have a culture which builds and encourages investment expertise. Where pools have chosen to build their own ACS structure, we should perhaps see higher costs as a badge of honour. But they will need more scale to bring costs down.


The pooling consultation is therefore not necessarily wrong to encourage scale, though in my view DLUHC should be more open to other solutions. I would really like to see them focus on an absolute cost target of, say, 40bps. That would then provide a cost budget for pools to allocate as they see fit. It might also incentivise the building up of internal expertise over time in higher fee asset classes such as private equity.



* For the detail here, please see ‘A study of global best practice in government funds: the opportunity for the LGPS’, Schroders 2018.


Comments


bottom of page