Blog 3 from Con Keating on the Proposed CDC Code
The CDC Code of practice has much to say about investment; much of which is far more intrusive than single lines requiring that assets held as part of reserves be investment grade, far more also than PSA 2021 or the CDC Regulations. The legislation for CDC, in the main, is concerned with the sufficiency of scheme assets relative to projected liabilities. The proposed Code of Practice however, goes far further making demands which effectively constrict a scheme to the Pensions Regulator’s view as to how investment management should be conducted. There is no legislative basis for this, nor should there be.
Among the items that are listed as matters more likely to satisfy the Regulator is:
“The trustees, on an ongoing basis: can show a logical connection between the trustees’ declared investment philosophy and the investment strategy being pursued”.
The presence of this demand really shows up the the lack of understanding of TPR about the realities of investment management. The classic caution in this regard is hardly new, and is due to John Maynard Keynes:
“There is nothing so disastrous as a rational investment policy in an irrational world.”
I cannot help thinking that I should pose a variant of this question to TPR:
Can you show a logical connection between this Code and the provision of high-quality CDC pensions?
Under that investment governance rubric, we also have:
“The trustees, on an ongoing basis: can show a thorough understanding of the investment strategy from the perspective of risk, return, markets, asset class and outcome”.
What purpose is served by viewing the investment strategy in all of these manners? What purpose is served by looking at the strategy from the perspective of markets and asset class(es) when risk and return have already been considered?
The Code considers requirements for data:
“There is sufficiently detailed information to enable trustees to monitor investments effectively, including:”
which is fine until we get to:
“…details of the investment returns and the risk profile of the investment strategy, which are consistent with their stated objectives”
So there is to be continuous second guessing of the investment strategy, which of course will almost surely lead to expensive misconceived interventions and abject failure with respect to those cherished long-term objectives.
“…a suitable range of risk metrics, which trustees use to inform their ongoing assessment of returns”
I think it is incumbent on the Regulator to precisely specify the metrics they would like to see here and how they may be used in an ongoing assessment. This requirement runs the risk of encouraging trustees to extend their exposures when volatility is low and to reduce exposures when volatility is high – a classic method of value destruction for savers and the cause of more than a few financial crises.
“…detailed performance attribution analysis of the investment strategy, including benchmark and peer group relative performance”
The requirement is not that trustees should perform attribution analyses, merely that they have the data to do so if they wish. Quite why they should wish to perform peer group relative performance analysis has no (overwhelming or other) supporting rationale, but it certainly involves the cost of collection of otherwise irrelevant information.
There is also a problem with performance attribution in that most approaches to this are time inconsistent – the long-term relevance of a factor is not necessarily the sum of its short-term constituents.
While still within the processes section of scheme governance, the consultation moves on to set requirements (as matters more likely to satisfy the Regulator) for risk management and introduce a requirement to maintain a risk register:
“There is a risk register to support the ongoing monitoring of risks and it has been considered and agreed.”
and this is a substantial management tool, as can be seen from:
“The risk register is regularly reviewed in detail by trustees, with considerations and decisions being documented and ownership and actions attributed, along with timelines for delivery.”
The idea that risk can be in any meaningful sense agreed by trustees is a fundamental misunderstanding of the nature of risk, and particularly so for financial risk. However, the obligation is no mere box-ticking exercise; it will be extremely costly as can be seen from the requirement:
“Appropriately skilled individuals are responsible for monitoring risk against the work planned and aims and objectives, with access to the management information and intelligence they need to carry out this task properly.”
That requirement alone looks to carry a cost of £500k – £1 million.
The risk register makes one other appearance in the Code in connection with the reporting of significant events. Under the rubric:
“A failure of the systems and processes used in running the scheme, which has a significant adverse effect on the security or quality of data or on service delivery”
we are told:
“In general, notifications of this event should look at divergence from the objectives set for the scheme and tolerance of risk as set out in the risk register.” [Emphasis added]
So the risk register needs also to record and list the trustees’ tolerance of the listed risks. We are offered the following narrative for this:
“This is important in considering the balance of probabilities that the failure has, or will have, an adverse effect that a scheme’s existing systems and processes cannot rectify without impacting on the security or quality of data, or on service delivery.”
This goes far beyond the Regulations which stipulate simply:
“(k) a failure of the systems or processes used in running the scheme which has a significant adverse effect on the security or quality of data or on service delivery.”
It is clear that TPR has a very particular and very narrow view of investment management – one which is dominated by asset prices and their short-term fluctuations, much like the accountants and their view of the world. Whereas, the problem of pensions is better framed as an income and expense exercise rather than a point in time asset to liability comparison.
The Code introduces an explicit ALM requirement under
“Testing and modelling in respect of the scheme”: “ We expect the modelling and testing undertaken and provided to the trustees to include asset liability modelling (ALM) alongside an explanation of assumptions and inputs.”
but qualifies this with the subsequent:
“We do not expect new ALM every year. New ALM should be undertaken where necessary to provide outputs that can be relied on to inform the necessary decisions.”
The following narrative is offered in support:
“This should provide assurance that outcomes are consistent with what is communicated to members, as well as providing trustees with a measurable way of understanding the variability of outcomes, to support trustee decisions.”
This is seriously misconceived; ALM does not measure outcomes. It relates the current market value of assets to the present value, derived using an arbitrary discount rate, of projected but highly uncertain future benefits payments. The variability of ALM outputs of course requires multiple iterations of the model with variation of the parameters that serve as inputs. Put another way, ALM used in this way can inform us about the variability of its model output, but it tells us nothing about its accuracy.
That narrative section also includes:
“ If modelling is based on an existing known model, then any changes to the underlying model should be explained.”
Quite why this was included is a mystery (some earlier burnt fingers, perhaps?) and there is more than a touch of the ‘kitchen-sink’ about it. But that is the defining characteristic of this proposed Code.