Decoding pension scheme benefit design

Estimated reading time: 8 minutes

This blog looks at a better framework for judging benefit design, stripping back the many decades of perceived wisdom within the UK and starting again from first principles – and looks at how popular options at the moment measure up.

Fundamentally, benefit design doesn’t change the fact that if you put the same amount in, you get the same amount out, overall.

The only thing that affects the relationship between what you put in and take out is the investment return received (I am assuming here that you want the money out at the same time – if you leave it in for longer the outcome will change, but that is true across all benefit designs). There is a link between benefit design and investment return but that link is another issue that can cloud the thinking around benefit design, so for now let us assume that investment strategy (and hence return) is the same under all design options. (We’ll come back to that assumption later, when looking at different scheme designs.

"The real power of benefit design is getting good value for money for the amount put in. The overall amount taken out is the same, but how is that money used – does it all give value, or is some “wasted”?"

Benefit design can potentially have a large impact on:

  • how the overall “pot” of money is allocated to individuals. (Do they all get enough, or do some get more than their fair share?) And
  • how is an individual’s allocation then used to provide the benefits. (Can it be targeted to provide the benefits they actually need?)

And, as we all know well, benefit design affects the risk. The risk comes from the fact that the relationship between the amount in and the amount out is unknown in advance. It will be the same across benefit designs, but we’re not sure how much you get out for each £1 input. Benefit design affects what happens when our expectations about the relationship change. If we think pensions have become more expensive do we:

  • fix the “amount out” (define the benefit), increasing the amount in, so increasing the contributions required (usually funded by the employer, so it is their risk), or
  • fix the “amount in” (define the contribution), reducing the amount out (so members’ bear the risk, and get lower benefits)?

Overall, this gives me 3 features to consider when thinking about benefit structures:

  1. Who bears the risk?
  2. How are resources allocated between individuals?
  3. How are resources targeted for each individual?

For example, DB schemes are very good at allocating resources to individuals (assuming the DB formula is the right one), and DC schemes are very bad at this (because what people get depends upon, for example, market conditions at the time they need the money). Conversely, DC schemes (since the pensions freedoms) give complete flexibility to target resources in the way members need in retirement, but DB schemes are very bad at this (unless you believe everyone wants guaranteed increases for life).

Comparing scheme designs

So, let’s think about some common scheme design options, and see how they measure up:

So, as is well documented, the employer has all the risk with DB schemes. As already noted there is very good targeting between members, but very inefficient targeting from an individual’s point of view. Around one-third of the overall cost paid goes to provide guaranteed pension increases – is that always the best use of resource? I cannot believe it is. Removing the statutory requirements for increasing pensions would help a bit, but even then, having to provide at least a fixed pension for life now seems pretty inflexible compared to what a DC scheme can offer.

A cash balance scheme is like a DB scheme before retirement, but instead of a promise of (say) 1/60th of salary for each year of service (which would give a pension of 50% of salary after 30 years), the promise is (say) a lump sum of (say) 1/3rd of salary for each year (giving a pot of 10x salary after 30 years) which can be used just as a DC pot can.

They haven’t really taken off in the UK, possibly because the DC freedoms didn’t exist when most people were moving away from DB. I think they are now worth a second look, because they score very well on both targeting of resource between individuals (a “DB” lump sum targeted at retirement), and also benefit from all the DC freedoms in terms of how benefits are drawn.

In terms of risk, they are one form of risk sharing – effectively the employer bears the risk pre-retirement, and the employee bears the risk post-retirement.

An aside on risk-sharing:

Cash balance is one form or risk-sharing. There are any number of ways to achieve risk sharing – the most obvious would be to offer a lower level DB benefit alongside a DC top up. This is actually a pretty sensible option – it gives a minimum level of protection for life, plus the ability to use the DC top-up to get the flexibility you want in retirement.

The member has complete flexibility to shape the benefits how they see fit (I am skating over the challenges involved in engaging with members so they can understand how to get the best from that flexibility, which can be done, but does need proper attention), but members bear the risk, and so DC schemes don’t do very well at ensuring that all members get their fair share – different market conditions will affect members differently – some will do better than others from this.

The traditional response to this problem is to reduce the investment risk taken to avoid the volatility that can affect members differently – but that causes other problems (i.e. lower expected outcomes), so is not ideal.

This means different things to different people, but in some form or other, it means adjusting the DC framework by pooling risks in some way to try to get better outcomes (in terms of how resources are allocated between members – for example, to allow more investment risk to be taken).

There are many ways this could be done, but here are a couple of more obvious examples:

  • Members could invest in a fund which pooled returns, trying to smooth our market fluctuations. (Just like a with-profits fund, in fact.)

  • A model that has been use in The Netherlands for many years is to have a notional, or target DB benefit that is then used to allocate assets between members – so resources are allocated in the same way as for a standard DB scheme, it’s just that if the scheme is “underfunded” (compared to the notional target), then everyone gets less.

Could the Dutch model work here?


It would need legislative change to make it possible, but even then I am not sure. The culture here is different. Dutch schemes always had the concept of benefits that were not guarantee – generally their DB schemes had discretionary increases, that didn’t always get paid.

Actually, what happened in The Netherlands seems to have been a replacement of DB schemes (with high levels of discretionary benefits) with CDC schemes which have a similar target benefit. This was driven primarily by a need to remove the volatility from balance sheets, rather than because the DB schemes were fundamentally too risky (remember those discretionary benefits – if only we had that option here).

Instead of providing a guarantee (which would put the liability on the balance sheet), Dutch firms seem to have increased the money paid into the schemes to provide the same level of (target) benefit – they replaced the guarantee with a buffer in terms of the funding.

So far we have talked about risk from the traditional point of view of either the employer has it, or the members do. Yes, it can be shared but only between those parties.

Actually, there is another option which could achieve similar benefits to CDC within the current DC framework – someone else deals with the risk. There are two main possibilities here:

  • Risks can be insured. The most obvious example here is members in a DC scheme buying an annuity. Perhaps not a good example, given the reputation annuities have, but I think they can have an important part to play, providing a minimum level of guaranteed income. A with-profits investment vehicle is another (probably also not very inspiring) example, but the point is the market is there for clever solutions insuring the right risks in an affordable way.

  • Pooling by fund managers, rather than by schemes. So rather than a CDC scheme taking responsibility for pooling, this could be done at the fund manager level, providing returns in some way linked to the experience of the whole population of investors in the fund. There are all sorts of possibilities here – sharing longevity risk via a manager pool, for example.

So, what would I do?

How about this as a framework, which would allow employers complete flexibility to design a structure reflecting their risk appetite?

What we used to call “pre-retirement”, but quite rightly we now reflect the fact that “retirement” isn’t likely to be a single date for people in the future. Basically while people are building up money in their scheme.

Some combination of cash balance and DC. This could be all cash balance (e.g. one-third of salary for each year worked); all DC; or a combination (say – one-sixth of salary up to £60,000, plus 10% of salary into a DC pot).

Any cash balance part would probably be based on average salary, rather than final salary.

Increases on any cash balance part could be discretionary (before retirement), to allow the scheme to weather poor market conditions, if required.

The relative levels of cash balance/discretionary benefits and DC would depend upon the employer’s risk appetite/desire to provide guarantees.

Work would be needed on the best investment options to help members manage any DC risk.

As a DC scheme – access full flexibilities – with lots of good engagement with members to help promote good outcomes.

Focus on finding good ways for members to access a minimum level of security in retirement – are there more cost effective options than annuities available? A very paternalistic employer might even decide to allow a part of the “pot” to be used to purchase a guaranteed pension within the scheme – i.e. retaining the DB-type post-retirement risk albeit at a lower level, with most of the pot being used in a DC way, accessing full flexibilities.