The federal government recently released exposure draft (ED) legislation that will introduce a retirement income covenant for superannuation trustees.
In brief, the ED requires trustees to formulate a strategy to assist beneficiaries to achieve and balance the objectives of:
- maximising expected income over the period of retirement;
- manage expected risks to the sustainability and stability of retirement income over the period of retirement; and
- to have flexible access to expected funds over the period to retirement.
A retirement strategy that optimises the achievement of these three objectives in a defined contribution environment has been the holy grail of financial advisors, retirement product providers and superannuation actuaries.
This article sets out a specific proposal to achieve all three objectives. It then follows a discussion as to how the specific solution can be generalised to a wider section of the retired population. Of course, even in a generalised form, the proposal will only be suitable for a segment of the retired population. One size does not fit all in retirement planning.
The design and implementation of the proposal are amenable to actuarial input.
Digression – late retirement
Although not part of the solution, it is worth noting that extending a person’s working life is a ‘silver bullet’ to achieving financial goals in retirement. An extra year’s employment provides an extra year’s investment return on peak balance, an extra year’s contributions and the expected period of funding reduces by a little under one year.
Maximise expected retirement income
To achieve the objective of maximising expected retirement income, the strategy is simply to invest in growth assets. The actual percentage in growth assets will depend upon risk preferences, but 70% plus is a reasonable guide, based on the default or MySuper settings of superannuation funds in Australia
For many retirees, retirement is for the long term. For a 65-year-old male, the expected lifetime is 22 years; for a female, it is 25 years (Australian Life Tables 2015-17, rated down three years). On both theoretical grounds and experience, growth assets are expected to provide the best return over these periods.
Explicitly rejected are such solutions as:
- mainly investing in defensive assets; the expected return is relatively low;
- using derivatives to dampen fluctuations; the underlying costs are high; and
- effecting annuities (other than investment-linked annuities); the asset allocation is oriented towards defensive investments; and the capital required to maintain the guarantees requires a return to the provider.
Invest in an account-based pension. For those whose balance exceeds the Transfer Balance Cap, invest the excess in an accumulation account.
This strategy provides complete flexibility since any amount can be withdrawn at any time.
Manage risks to sustainability and stability
There are four steps in the process:
- Set a funding period. This would typically be the period over which the retiree has a 25% probability (or whatever probability that the retiree is comfortable with) of surviving and/or the period, at the end of which it is thought likely that the retiree will have reached his/her healthy life expectancy and will move to a retirement home or aged care facility.
- Set aside an amount for contingencies and/or to accumulate to a lump sum at the end of the funding period. Whilst an income should be the primary form of benefit in retirement, a relatively small lump sum to meet unexpected medical or repair bills is one of the highest priorities for peace of mind in retirement.
- Calculate the monthly income that is expected to reduce the remaining balance to zero at the end of the funding period. Other than the initial balance and the funding period, the only inputs to this calculation are the expected rate of investment return and the expected rate of inflation.
- At the end of the first and subsequent years, repeat step 3 with the then actual balance and reduced funding period.
The premise of this methodology is that the retiree is satisfied that at any time, the income being drawn down is the ‘correct’ amount to last for the funding period. Of course, each year the ‘correct’ amount is recalculated. The retiree needs to have a mindset that is comfortable with the concept of homing in on a constantly moving target. At the moment it is likely that only more financially sophisticated retirees will fall into this category. However, it is worth exploring whether education and/or financial advice can assist a wider group of retirees to consider the proposed approach.
Another important issue is that the retiree is happy with his/her financial position at the end of the funding period. There is a considerable safety net here, in that the end of the funding period should be well into the frail period of retirement, where expenditure is relatively low. Provided that the retiree owns his/her home or accommodation in a retirement home or aged care facility, the level of income required more than the Age Pension should be quite modest.
A death benefit is automatically provided as the account balance at the time of death. For a retiree leaving a surviving spouse, this benefit has an appropriate shape, being relatively high if the retiree dies in the early years of retirement and being relatively low if the retiree dies in the later years of retirement.
As previously noted, the proposed method automatically achieves the two legs of the trilemma of maximising expected return and of flexibility. Provided the funding period and acceptable end resources are judiciously chosen, sustainability is also achieved.
There remains the issue of stability of income. Two parameters were chosen to assess stability: the first parameter is the variation in drawdown from one year to the next; the second parameter is the number of times that future drawdowns exceed the first year’s drawdown plus inflation, as measured by the increase in CPI.
Backtesting carried out
The below appendix sets out the parameters of a backtest, using actual fund data, covering 128 trials each with a funding period of 18 years. The results were:
Change in drawdown from previous year
Average number of instances
Cumulative number of instances
More than 10%
Zero to 10%
-5% to zero
-10% to -5%
-15% to -10%
-20% to -15%
Less than -20%
The extreme negative results all occurred due to the Global Financial Crisis (GFC), and all 128 trials included the period of the GFC.
If a drop in income from one year to the next of amounts up to 10% are deemed reasonable, then the proposal is successful 15.8 times out of 17.0. It is up to the retiree’s own risk preference as to whether this outcome is acceptable. Again, education and/or financial advice is essential.
Also, on average, for 11.2 years the drawdown was greater than the original drawdown plus CPI, and for 5.8 years it was less. This aspect is discussed further below under the headings ‘Sequencing risk’ and ‘net rate of return’.
Backtesting could be widened by using historic sector returns and combining them to construct plausible growth portfolios. See for example Grenfell CR, 2021, Australian Investment Performance 1959 to 2021 (and Investment Assumptions for Stochastic Models) Presented at the All-Actuaries Virtual Summit on 10 May 2021 (and was updated on 22 August 2021).
Backtesting has the advantage of bringing a ‘real life’ perspective to analysis; however, the number of scenarios is limited. The analysis could also be carried out using investment consultants’ asset simulation models. Many more scenarios could thereby be tested. As usual, it must be caveated that models are projections, not predictions.
Expanding the framework
As set out above, on average, 5.8 years out of 17 resulted in the actual drawdown being less than the original drawdown plus CPI. Closer inspection of the results shows that there were ‘successful’ periods in respect of this parameter (when there were three or less instances of actual drawdown being less than the original drawdown plus CPI), being periods commencing 31 January 1993 to 30 April 1997 and periods commencing 30 November 2002 to 31 August 2003. The ‘failure’ periods were periods commencing 31 May 1997 to 31 October 2002.
It is clear what is happening; in the first successful period, high returns in the early years were sufficient to offset the low returns during the tech wreck and the GFC. The failure period occurs when there is insufficient time to build up a buffer before the tech wreck and GFC. The second successful period occurs after the completion of the tech wreck where there is sufficient time to build up a buffer before the GFC.
This feature of outcomes being profoundly affected by the starting point is known as sequencing risk. Since no one ‘rings the bell’ when markets peak or bottom out, sequencing risk will always exist.
An approach to mitigating the effects of sequencing risk is to utilise the mean-reverting characteristic of markets; that is, a period of above-average returns is inevitably followed by below-average returns. Hence, if the five years (say) before retirement have had high returns (relative to the assumptions employed – see the Appendix) then the retiree’s expectations should be lowered. Further work needs to be done to quantify this concept (e.g., calculate the notional account balance if assumptions had been borne out in the five years leading up to retirement. If this is lower than the actual balance, the initial drawdown on the notional balance should be the benchmark for the CPI test).
The effect of the Age Pension has not been considered in the analysis. Fortuitously, the effect of the Age Pension will be to dampen volatility, through the operation of the Means Test. Incorporation of the Age Pension is an area of further investigation.
Net rate of return
In calculating the drawdown in the backtesting carried out, the rate of return was set at the Trustee’s objective (see Appendix). Anecdotally, Trustee Directors tend to set the objective so that there is a two-thirds probability of success. (Interestingly the CPI benchmark was met in practice 11.2 times out of 17 years, and two-thirds of 17 is 11.3).
Depending upon the retiree’s preferences, the net rate of return could be dialled up or down. A retiree who prefers immediate income could choose a net rate of return where the probability of exceeding CPI is 50%; a risk-averse retiree might choose a net rate of return with, say, an 80% probability of exceeding CPI.
Investment-linked annuities (either immediate or deferred) are very similar in concept to the proposal in this note, and both offer whole of lifetime income. Compared to the proposal, their disadvantage is that the provider will necessarily be conservative in setting future mortality assumptions and will be seeking a profit margin on capital required for the longevity guarantee. Thus, maximising income, and to certain extent flexibility, are reduced; albeit to a far less extent compared to annuities providing guarantees in respect of dollars or CPI linked.
Under the proposal, it would be permissible to draw down less than the calculated drawdown. The question then arises; to what extent could drawdowns be increased at a later stage to catch up, but not exceed, the original plan. The answer is to calculate a notional balance, being the balance that would exist if the original plan was followed. The maximum catch up drawdown is the amount that brings the actual balance down to the notional balance. The calculation of the notional balance is straightforward.
CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.