We’ve just released a research paper titled How to approach quantitative assessment of retirement income strategies, a companion piece to Assessing retirement income strategies… when outcomes are but a promise. A detailed summary of this paper is found here. This companion article stands back and considers the assessment challenge in broader terms.

A highlight of the research process was the feedback we received. Thanks to the many who engaged, most of who (but not all) are acknowledged within the paper. If anyone else wants to engage with us on future work, please make contact!

Below we draw out three main observations arising from the feedback. Note that this work focused on retirement income ‘solutions’ – how income is generated through an integrated investment and the drawdown strategy – rather than how those solutions are delivered to members through guidance, advice, and so on.

Read Part II of this series: How do you assess if a retirement solution is any good?

Role and use of models

Our paper was directed at how quant processes can assist to design and assess solutions. While everyone will have their own spin on this matter, we offer three reflections:

  1. Groups involved in solution design may need more detailed models than those who assess these strategies such as fund boards, research houses, and potentially APRA. Design is more concerned with the details, while assessment focuses on the broader outcomes.
  2. Fund members require solutions to be presented in a completely different frame. While models may provide the common engine, the output information shown on the dashboard needs to be tailored for the audience.
  3. While models might be used to find ‘optimal’ solutions, this can be dangerous if applied naively. Optimisation is technically difficult in retirement due to many ‘choice’ variables, notably around investments and drawdowns. Reliability of optimised solutions is diminished by uncertainty around assumptions. Optimisation misses a bevy of more practical considerations like cost, feasibility and likely member acceptance. In our view, finding a good – or better – practical solution is the aim of the game.

Mixed objectives at play

Multiple possible objectives could be adopted in designing a solution. The Retirement Income Covenant (RIC) provides broad objectives around expected income, income risk and flexible access to funds. These fall short of specific objectives against which model outputs can be compared.

Our paper highlights the importance of this issue by presenting case studies exploring two different objectives – income optimisation and income targeting. The feedback process revealed that funds are indeed using different objectives. Some are working towards an income target such as ASFA comfortable; whereas others are aiming to maximise but manage income risk and thus delivering income variability in the process.

Also evident was significant uncertainty and variation around how access to funds is viewed. Unanswered questions include how much access to funds is required, how it might be measured (dollars versus percentage; over time or at a point in time), and how it is calculated (e.g. should other financial assets and home equity release be considered?).

The importance of confidence to spend appropriately in retirement received a few mentions as another objective, perhaps to be assessed through member testing. Food for thought!

Lack of clarity in crucial areas

We have already raised a lack of clarity over the objectives for income and access to funds. Some important modelling assumptions for solution design were also a source of debate. Two matters are worth highlighting:

  1. Assumptions around the indexation of the Age Pension and member consumption were raised as crucial. This is not a new issue, having been a source of strong debate around findings of the Retirement Income Review and the Grattan Institute. To explain: (1) the Age Pension might be indexed against wage growth in accordance with the existing rules, or inflation due to concern over whether wage indexation is fiscally maintainable; and (2) consumption through retirement might be either indexed at inflation to reflect maintaining real consumption, or at wages to support “keeping up with society”.
  2. Another debated issue is whether ‘expected income’ should be mortality weighted. To explain through example, say a solution delivers no income additional to the Age Pension at age 104. Should this outcome be down-weighted because the likelihood of living to 104 is very low? Or should consumption at any age be considered equally important at all ages regardless of likelihood of reaching each age?

The stance taken on these matters impacts on the need to have some longevity protection within a solution, either through adding a lifetime income stream or limiting drawdowns so that the assets last. Commenters had quite different views on these issues.

What comes of all this?

We observed that many funds are engaged with the process of how to quantitatively assess their retirement income strategies. This will contribute to them developing better quality strategies.

Beyond this, we saw a large degree of diversity in both the solutions being developed by funds and the way they are being assessed. At a system-level this dispersion risks complexity, inhibited understandability and diminished comparability.

We found ourselves wondering whether complementing the principles-based RIC with further clarity of assumptions and expectations in some key areas would provide clearer direction for funds and improve outcomes for retirees.

David Bell is executive director at The Conexus Institute
Geoff Warren is an associate professor at ANU and research director at The Conexus Institute

One comment on “Significant challenges in assessing retirement income strategies”
    Gishta Awmee

    Interesting read.

Join the discussion