This detailed paper by the OECD RiskLab makes a useful contribution to the knowledge on DC pension guarantees and their feasibility.

It rightly distinguishes between a number of different guarantees that might seem superficially similar, but could differ greatly in value (and cost) :

1. G_0 : a guarantee * at point of retirement only*, that pot will be worth at least as much as all contributions paid in over the member’s working life

Given that most assumptions about future investment returns will mean that, over a 30 year+ working life the total Net Asset Value (“NAV”) of a member’s pot will be expected to greatly exceed the contributions paid in this guarantee is shown to be (optically) quite cheap, costing around 1% of the value of contributions paid in. So far, so good.

However, there are several practical points which mean this approach probably won’t work. Firstly, practical considerations dictate that any pension pot should be **portable
**to another provider in the case of an employee changing jobs. This is logistically difficult in the case of a guarantee over such a long period of time. Secondly, capital losses will not necessarily be prevented in the early years hence it may not in practice satisfy the objective of increasing confidence in the savings system.

These two considerations motivate a new type of guarantee :

2. G_Ongoing : a guarantee **valid at a number of future points through time (eg annual)**** **that the member will get back at least the value of contributions invested. This covers portability aspects and also gives more certainty in the early years, however it is of course much more costly to provide, potentially costing in excess of 10% of the value of the contributions, according to the paper, and at the point of retirement doesn’t actually guarantee a higher amount than (1), just more flexibility in the interim.

Other types of guarantee focus back on the point of retirement only, but seek to provide a guaranteed level of investment return on the contributions invested, as opposed to simply guaranteeing the sum contributed.

3. G_2 and G_4 guarantee respectively fixed levels of 2% and 4% annual investment growth. According to the paper the cost of guarnateeting 4% p.a. is roughly 4 times the cost of guaranteeing 2% p.a. However a guarantee of even a fixed level of investment return on the contributions invested doesn’t protect the investor against rising inflation, and may not seem as attractive as initially thought if a period of high interest rates occurs, this motivates two further sorts of guarantee:

4. G_real is a guarantee that the contributions will grow at least at the rate of inflation over time. The cost of this is similar to G_2 although slightly more expensive partly because the rate of future inflation is not know with certainty in advance.

5. G_Float is a guarantee that references the prevailing 1 year market rate of interest at each point in time and guarantees returns at this level. This protects an investor against dramatically rising interest rates through time, however the paper shows this to be the most expensive.

I would suggest another “family” of guarantees be considered, which focus not just on protecting contributions, but on the preserving some or all of an investors assets in any given year. For example, a floor on the losses an investor could suffer in any give year of say 15%. While this doesn’t explicitly protect the contributions made, it provides more comfort to encourage saving in the early years. In the later years, if significant investment returns have occurred it is also potentially more valuable and useful than a guarantee that references contributions only.