This is a longer version of the article published in the Local Government Chronicle on 6th September.
I’d better start this with a disclaimer of sorts. I expect this article to upset a whole swathe of people. But the future of pensions, and for me the Local Government Pension Scheme (“LGPS”) in particular, is too important for too many people to get it wrong. And there is a substantial risk that is what is about to happen.
Is the LGPS really inefficient?
The Chancellor has announced a Pensions Review and included the LGPS in the first stage. Her objectives can be summarised as threefold: consolidation into larger scale entities; better value for money; and for pension funds to play a greater part in funding the investment into society which the Government wishes to make. The first two are connected; the third may be.
She believes that the LGPS as currently managed is inefficient. Her starting point is reasonable – the proverbial woman from Mars wouldn’t start with 86 different teams to run what is ultimately a single fund. But is the outcome so inefficient?
LGPS costs are lower than the Maple 8
The Chancellor uses the Canadian Pension ‘Maple 8’ system as a model. Data from 2006-2015[1] gave average investment costs across the Canadian system of 48bps against a comparator group of 50bps. And the LGPS? It’s not an exact comparison, but 33.5bps in 2017 and 49bps in 2023 on the publicly released data. The Scheme Advisory Board report gives a figure of 43bps for 2023.
The same study found that the Canadian system in-sourced implementation much more. This is one major difference from the LGPS today, where the direction of travel over the past twenty years has been towards outsourcing. Even the pools tend to use external managers.
Even despite this, the overall cost as a % of assets is similar. Readers could infer from this that there is scope for the LGPS to reduce costs further by insourcing more; or alternatively that there is less scope for cost saving because the LGPS already pays competitive rates for its external managers. But the study does not make a case that the LGPS’s current structure is inefficient.
LGPS Value Added is comparable
What about the value added by moving away from the risk-free (i.e., matched to liabilities) portfolio? This measures the outcome of both the choice of risk assets and the returns over time. The study found that over the ten years Maple 8 generated 2.2% of extra return from taking 12% of liability mismatch risk. This was superior performance to the U.S. or Dutch pension funds in their universe.
To compare this with the LGPS, I used data from the recently published Section 13 Government Actuarial Dept report with data to March 31st 2022. To create a liability return target (i.e., the equivalent of the Study’s liability portfolio return) I used the average discount rate. However, the GAD 13 report does not give this directly, so I backed it out from the risk-free rate of 1.9% (i.e., the nominal 20 yr gilt spot rate) and added their average implied annual asset allocation outperformance number. The average is about 2.6%[2], with over half of funds between 2.1% and 2.8% regardless of the actuarial methodology used. That gives a notional liability return target of between 4% and 4.7%.
Annualised investment returns were 7% annualised over the ten years to March 2023 according to the Scheme Advisory Board’s report. If we deduct 49bps of costs, that would suggest a net return of 6.5% or net value added of just over 2%.
These calculations come with many caveats[3], but my point is that this is not out of line with the 2.2% of extra return generated by the Maple 8. Again, the case that the outcome under the Canadian Maple 8 system is superior is not proven.
Should the focus really be on funding the unfunded schemes?
Another key comparator of efficiency is the employer contribution rate compared to unfunded public sector schemes. As the latter do not have any financial benefit from investment returns, it is no surprise that their employer contribution rates are higher. The Civil Service Pension Scheme and Teachers rate is 29%, the NHS 24% and the Armed Forces 71%[4]. In comparison most LGPS employer contributions are between 15% and 25%, averaging 21%.
If the Government really wishes to reduce the overall cost of public sector pension funds, perhaps it should start funding the currently unfunded public sector schemes? Not only would it build up a large pot of funding, but in the longer term the cost of providing these pensions would reduce as investments generate returns.
Independent governance is the key to Maple 8
Another major feature of Maple 8 is independent governance. That means a Board which is completely independent of its employers and management. The reason this is so important is because employers can hold the Board to account without conflict of interest, while the Board can similarly hold management to account, and if necessary change them.
The current LGPS structure is very different. The fact that the scheme administrator is the same body as the largest employer makes independent governance difficult. The Good Governance recommendations do their best to compensate by recommending a nominated Head of Pensions, but it cannot remove the inherent conflict of interest. And they have not yet been implemented into law or guidance.
Almost all the pools have partner fund representatives on their Board, partly because their owners mostly set them up using the Teckal exemption and therefore had to demonstrate control. I have elsewhere argued that when the new post-BREXIT public sector procurement rules come into law in October this will no longer be necessary[5]. Unlike the Canadian funds, the pools also almost all have management representatives on their Boards, which I would argue precludes independence.
Would the outcome of a Canadian model be more investment in the UK?
And let’s momentarily imagine a hypothetical world where the LGPS was converted into a single professional team of managers and a truly independent board, and the Government actually paid the employer contributions of unfunded schemes into a small number of similar funds. With a nod to George Osborne’s “half a dozen” pools, we could call it Brit 6. To reach this point, the Government would need to cut a swathe through the LGA 1972 Act for starters, and having to invest their contributions rather than pay them out directly would wreak havoc with their budgets.
Brit 6 might give better value for money, though I don’t think the Canadian example makes that case clearly. But would the Chancellor actually get more funding for her investment projects? I think it is highly unlikely, as a professional management would be far more aware of its fiduciary duty and less vulnerable to pressure. One important element of the Canadian Scheme is a governance structure which makes it more difficult for politicians to meddle.
As just one example, according to a recent article in Pensions and Investment the Canadian funds only invest 7% of their infrastructure allocation in Canada. The equivalent domestic percentage for the LGPS is 67%. Perhaps it’s a case of the Chancellor being careful what she wishes for.
---
[1] CEM Benchmarking data used by Keith Ambachtsheer ‘The Canadian Pension Model: past, present, and future’ Journal of Portfolio Managemement, April 2021.
[2] 2022 S13 report Appendix Table 5.1.
[3] The main ones are that my LGPS calculations take no account of the level of liability mismatch risk, and that the ten year periods used are significantly different (2006-2015 and 2013-2022).
[4] Armed Forces personnel make no employee contribution, so the comparison is not exact.
[5] ‘A New Model for the LGPS Pools?’, Linchpin article in February 2024.
Comments