Straight to content

DC: Crossing the river to fetch water?

‘A look ahead to 2023’ download our full report below

Next year we will hopefully see the Department for Work & Pensions’ (DWP) eagerly anticipated Collective Defined Contribution (CDC) consultation paper. There has already been a lot of debate this year around the benefits and challenges. But most claims in this debate are anchored to models, and those models rest on strong assumptions about the future dynamics of the financial markets.

A model is a good servant, a bad master and a terrible religion

Pooling risks that are diversifiable, such as individual longevity risk, makes perfect sense, but pooling of systemic risks, such as financial market risk, doesn’t. Some believe in time diversification, that stocks are less risky over time, and make use of that as an argument for
sharing financial risks across generations. The shadier versions of CDC are built on this belief.

The problem is that the ‘evidence’ backing time diversification is derived from the mean-variance framework. But that ‘evidence’ holds if, and only if, the model holds. Nobel Laureate, Paul Samuelson , Professor Zvi Bodie, and many more academics, have shown that time diversification does not hold. It is an empirical observation but should not be taken for granted.

We should not anchor financial risk sharing solutions on models that don’t capture the dynamics of the real world. In the UK, the recent move in 30yr Gilt yields, from 0% to
+2% in a week was so unlikely, that according to most traditional models it probably would never have happened. The Japanese crash in the 1990s followed by flatlined real stock returns are even more unlikely, according to traditional models. To paraphrase Amory Lovins, an
American energy policy analyst, a model makes a good servant but a bad master and a worse religion.

The North Star for CDC design

The north star for any funded pension design, is that future generations should not pay for current generations. The generational accounting principles, proposed by Professor Larry Kotlikoff, as a way to measure the impact on future generations, should be used as a litmus test for any CDC design.

A good way to achieve this is to separate accumulation and decumulation in the pension design. Keith Ambachtsheer, a world renown thinker on pension design, argues for the two pot model based on the difference in objectives; solving a savings or a spending problem.

For active members, the economic value of pooling individual longevity is limited. From a family perspective, the risk in the savings phase is to die prematurely. In that case it feels better to leave the pension savings behind for the family rather than to share it with other members
of the pension fund. Therefore, a CDC design is not needed for the savings phase.

For the spending phase, CDC solutions provide an alternative to traditional annuities. Well-designed CDC solutions periodically rebase pension rights based on realised investment returns and mortality gains. Good solutions do not include smoothening or buffers motivated by the vague arguments of time diversification.

Similar but different

Similar outcomes can be achieved by allowing for longevity pooling within an existing individual DC solution, such as a Master Trust. This has been done successfully in both Australia, Canada and the Netherlands. In terms of outcomes, a well-designed CDC solution and an individual DC solution with longevity pooling will be similar. The main difference is to be found in the accounting approach – pension rights or units – which is largely a matter of taste.

In anticipation of a CDC consultation paper next year, I can’t help but think; are we not crossing the river to fetch water?

Stefan Lundbergh, Head of DC Design

And is it not perhaps a simpler and more practical solution to allow Master Trusts pooling individual longevity within a dedicated section?

A look ahead to 2023

A year of reflection, consolidation and strategies.

A look ahead to 2023

"*" indicates required fields

This field is for validation purposes and should be left unchanged.