Life Cycle Funds
Saving for Retirement
The purpose of retirement saving is ultimately to provide a degree of financial security in retirement. The extent to which this financial security can be provided is a function of:
- Total contributions made during the working career (which is a function of contribution rate and length of career)
- The level of investment earnings on those contributions (in both accumulation and drawdown phases)
- The length of time in retirement to be funded from the accumulated contributions
- The level of funding (drawdown) in retirement.
In the process of accumulating sufficient funds to facilitate an appropriate outcome for the post-retirement phase the key objective which the investment industry can influence is the second one – the level of investment earnings on contributions. Considerable debate has taken place around the appropriate model for pre-retirement investment, some of which has also spilled over into the MySuper discussions.
The primary objective of the accumulation process should be to provide an income stream in retirement. Consequently, the objective of the investment component of this will be to provide the maximum income stream in retirement for as long as possible with greatest certainty. The question for our industry is therefore which investment policy (or series of policies) should be adopted to achieve this objective.
Intuitively the above objective should mean that the “riskiness” of the investment policy should reduce as the requirement for certainty grows (i.e. as retirement is approached) and then when in retirement less risky investment approaches are adopted such that the income stream certainty is increased. We note that some level of risk is often necessary and justified in order to increase returns as the zero risk option (i.e. absolute certainty) is generally too expensive.
In Australia the default investment option for pre-retirement has been a “balanced” or “growth” type fund - which has meant relatively fixed growth asset exposure of circa 70% and a real return objective of circa 3.5%-5.5% p.a. This may be scaled back slightly to an option with lower growth asset exposure (as a result of advice or in some funds’ cases automatically) in the final few years before retirement. This balanced investment choice is typically the default for all members (or at least those with more than a few years to go until retirement).
Once retirement is reached, more conservative options may be adopted either by having slightly less exposure to growth assets and/or reserving a number of years (say 2-5) of expected future cash flow into cash.
However, it has generally been accepted that maintaining some exposure (often a considerable amount) to growth type assets in retirement is still warranted as life expectancy at retirement is still measured in decades.
As noted, while in some funds age specific default options are being used, predominantly the “balanced” option is used by many members (and in many cases post-retirement members as well). The issue that has become more pertinent in recent years (as a result of market volatility) is that as members approach retirement their degree of tolerance for “risk” is reduced. As such, as retirement nears the investment strategy should “de-risk” to provide greater certainty of retirement outcomes.
In particular, one of the solutions that has been proposed to this problem has been the use of Life Cycle or Target Date funds. In a Life Cycle fund the asset allocation of the fund starts with a greater allocation to growth assets and then reduces slowly as the member approaches retirement. Consequently in the earlier years there is considerable exposure to growth assets while in later years the allocation to defensive assets increases. Target Date funds operate in a similar fashion with a specific fund being set up based on the likely retirement date for the individual.
Intuitively Life Cycle funds seem like a great solution in that they automatically “de-risk” the investment portfolio as retirement approaches, and in some markets like the United States they have received considerable cashflows. Indeed, even the U.S. Department of Labor endorsed Target Date funds in the Pension Protection Act of 2006 when it designated them as "qualified default investment alternatives" for all 401(k) plans. According to S&P who now produce a Target Date fund index, more than $200 billion had been committed to these funds by early 2008.
However, while intuitively the Life Cycle strategy sounds appealing, the simplicity of the approach masks a number of problems.
What is “risk”?
Again, the purpose of the investment policy is to provide the maximum income stream in retirement for as long as possible with greatest certainty. Consequently, risk in this case is the risk of the income stream being less than required or the capital value falling in retirement such that future drawdowns cannot be made (or only at a reduced rate).
At this point, we should also remember that for many members prior to retirement the option exists to either increase contributions or continue in some form of employment. However, once retirement has occurred, while not impossible, this becomes significantly more difficult. As such, we would contend that at the point of retirement there is a marked shift downwards in risk tolerance. Consequently, the tolerance for risk probably looks more like the chart below. The extent of the downward shift in risk tolerance on retirement will be a function of the propensity (desire and availability) to seek suitable employment.
If risk is about maximisation of income streams in retirement we should consider how best to achieve this in terms of investment policy. Central to this is that the measure of “risk” which funds are aiming to “de-risk” is generally assumed to be based on the asset allocation. That is more growth assets = more risk, less growth assets = less risk. While much of our industry’s marketing literature is based around this premise, it is nonetheless a large assumption (and in our view wrong).
Path Dependency and Volatility
The nature of the accumulation and decumulation process is such that the capital value on which returns are earned is not static. However, the process by which our industry assumes, models, measures and reports investment returns is based on time-weighted returns (i.e. a traditional geometric return). Time weighted returns assume the capital value is static. This is very useful in the context of considering an aggregation of members and measuring say the performance of a particular strategy over time, however it is not that useful for individual members as individuals earn money-weighted returns.
Consider the example of a fund that “earns” a return of 30% in year 1, 0% in year 2 and -10% in year 3. This would be represented as an annualised return of 5.37% p.a. If we now think of an individual who had $100 in the fund at the start of year 1 and contributed $100 in year 2 and another $100 in year 3, the end value is only $297. Had this order of returns been reversed, the end value would be $377 - very different outcomes (and a long way from the reported annualised return).
More generically, consider the example below for an individual who contributes 9% of salary from age 20 until retirement at age 60. In scenario A the earning rate is 10% p.a. for 20 years followed by 5% p.a. for 20 years. In scenario B this series of returns is reversed. The average annualised return is 5.8% p.a. The chart below shows the accumulated value (as a multiple of the salary which increases with inflation at 3% p.a.).
Clearly the differences are major – in this case 130% more at age 60 under scenario B.
A similar effect occurs in post-retirement. Again, to illustrate, consider the example shown in the chart below of an individual in drawdown for 20 years with a starting balance of $600,000 who withdraws $40,000 p.a. indexed at 2% p.a. In scenario A the investor earns 8% p.a. for 10 years then 4% p.a. while scenario B the return series is reversed. The average annualised return of both scenarios is the same at 5.98%p.a., however the end result for the investor under each scenario is very different given the drawdown that occurs.
In reality, given the path of wealth accumulation and decumulation, a better representation of the risk profile through time is characterised below.
To further illustrate this point, the table below shows the effective increase in benefit as a multiple of final salary after 40 years of contributions from a 1% increase in the contribution rate or the earning rate in the first and last 20 years of accumulation.
The table shows that a 1% increase in contribution rate in the first 20 years will have broadly the same effect as a 1% increase in the investment earning rate in the first 20 years of accumulation. There should be no question that the 1% increase in contribution rate is far easier to achieve than an additional 1% p.a. over 20 years! However, in the last 20 years the situation is considerably different.
Life Cycle Funds
We have shown above mathematically that:
- In the early years contribution rates have a far greater impact on the end benefit than earning rates, however in the later years earning rates are significantly more important; and
- The order in which returns are earned is extremely important to the final amount – i.e. absolute volatility of earning rates is as important (or more important) than the average earning rate.
The concept of Life Cycle funds is such that the exposure to growth assets – as a (imperfect) proxy for risk – is high in the early years yet drops away during the later years. While different Life Cycle funds take different approaches, we could generalise for the majority that in the early years the exposure to growth assets is greater than that of many traditional default options while in the later years it is somewhat lower. This is clearly not that beneficial in light of the first point above.
In respect of the second point Life Cycle funds fail (along with traditional balanced options) in that their proxy for risk is the exposure to growth assets. We have shown in other papers how the exposure to growth assets is not a good representation of risk – in this case absolute volatility. More technically, if the purpose of the accumulation is to generate a real income stream in retirement the aim should be to minimise the volatility of this potential real income stream. That is not a function of growth asset exposure (or indeed defensive asset exposure where the defensive asset is benchmarked with circa 3.5 years duration and against nominal interest rates).
Any investment policy whereby the asset allocation is fixed in some pre-determined fashion with no reference to underlying valuation, expected return and downside risk over the short to medium term runs the significant risk of generating substantial absolute volatility and not meeting the individuals’ investment objectives irrespective of where in the life cycle spectrum it falls.
The chart below shows for example the substantial absolute volatility of rolling 3 year real returns of typical default balanced options over the last 25 years.
Bearing in mind the substantial volatility shown above, consider the impact of an individual (un)lucky enough to retire (or near retire) in 1987, 2002 or anytime in the last 3 years. While this is generally part of the sales pitch for Life Cycle funds the reality (as we saw explicitly with the US experience) is that the small reduction in growth assets exposure wasn’t enough to properly insulate those close to retirement from the market volatility.
Longer term evidence
To more robustly test our thesis that Life Cycle funds are unlikely to generate better investment outcomes for members we have conducted a historical test of a sample Life Cycle fund vs a sample balanced fund. Given the significant time horizon required for a historical test (where we need circa 40 years of data for one complete result) we have constructed a basic Life Cycle fund and balanced fund assuming an investment universe of Australian equities, Australian bonds, International equities and Australian cash.
Our assumed balanced fund has an asset allocation of 70% equity, 30% defensive. The Life Cycle fund assumes an initial allocation of 90% equity/10% defensive that changes as per the chart below.
Examining the data since 1900 we can calculate the average annualised return differential between the Balanced fund and the Life Cycle option based on the year of commencement.
In analysing the differences we break the data into two periods – commencement dates for accumulation from 1900 to 1971 and 1972 to 1999. We do this as for an individual to retire after 40 years of accumulation by the end of 2010 they would have to have commenced employment by 1971. Consequently the start dates 1900 to 1971 represent the full 40 year accumulation period while the dates after that represent only a partial accumulation period. We do not consider individuals who commenced employment in the last 10 years for the analysis as the time period is relatively short.
From the table above it appears as if the Life Cycle option is relatively the better option - outperforming the balanced fund 73% of the time (using the full century of results).
However, the average return differential was negligible across the century. The year by year results are shown in the chart below.
If we now conduct the same analysis but account for the accumulation process over the members working career and consider the eventual end accumulated benefit, a somewhat different picture emerges.
We can see from the table above, the nature of the accumulation process is such that the balanced fund (which is far from the perfect approach) produces generally better outcomes than the Life Cycle option after we take account of the pattern of retirement saving. More importantly, where the Life Cycle fund does come out better, the results are at worst only 8.2% less than the equivalent balanced fund result. On average, the balanced fund strategy resulted in an end benefit nearly 12% higher than the Life Cycle fund. The chart below shows the outcomes year by year.
Only nine times since 1957 has the balanced fund fared worse than the Life Cycle option and even then by only 0.8% on average. Interestingly the skew is quite clearly to the balanced fund delivering better outcomes than Life Cycle fund. We would note at this point however that the final results were quite sensitive to the initial Life Cycle fund and the balanced fund asset allocations chosen, albeit generally the balanced fund option still did comparatively better than the Life Cycle fund once accumulations were taken into account.
As noted previously, there has been considerable take-up in the US of Target Date/Life Cycle type funds – partly as a result of a quasi-Government endorsement (word of warning MySuper) and (no doubt) partly a result of the marketing “attractions” for 401(k) plans. However, Target-Date funds performed so poorly in 2008 that in many forums their premise was called into question. Portfolios geared to members retiring in 2010 fell by 25% on average. Similar funds offered by different firms turned in significantly different results: for example one fund offered by a major US bank for retirees in 2010, lost 11% in 2008 (still not a great result), while another plunged 41%.
As a result there has been some debate and analysis on the suitability or otherwise of these plans for members. While the US situation is generally different to that of Australia in that the plans offered are retail in nature and may charge large initial commissions of 5-6% and substantial ongoing expense ratios compared to the more institutional nature of potential Australian offerings, some comment on the academic work that has been conducted is still useful.
Interestingly, one of the first academic analyses of the topic was a seminal paper by Paul Samuelson1 in 1969 titled “Lifetime Portfolio Selection by Dynamic Stochastic Programming”. That paper and another by Robert Merton2 in the same year effectively argued from theory that in the absence of labour income the optimal life cycle portfolio would NOT change the allocation to stocks through time.
Since then considerable research has been conducted into the topic – particularly with respect to the inclusion of human capital and in the face of uncertain return distributions across multiple time periods.
More recently, Basu and Drew (2009)3 and Shiller (2005)4 come up with similar results to our own:
“the size of the portfolio at different stages of the lifecycle exerts substantial influence on investment outcomes and, therefore, should be carefully considered when making asset allocation decisions. The evidence presented in this article lends support to the view espoused by Shiller (2005) that the growing size of the plan participant’s contributions in later years calls for aggressive asset allocation—quite the opposite of the strategy currently followed by lifecycle asset allocation funds.” [Basu and Drew (2009)]
From Spitzer and Singh (2008)5: “Target-date funds are marketed for their management ease and their perceived safety in the long run; those who elect target-date funds in retirement believe that decreasing equity exposure in retirement provides more income certainty. The findings here suggest that target-date funds are subject to a higher shortfall (longevity) risk and do not provide larger estates. Investment companies currently offering such funds should rethink the asset allocation strategy of these funds.”
The objective of the investment policy in a defined contribution accumulation fund is to maximise the value of the income stream in retirement with a given (increasing) degree of certainty. As such, it is widely recognised that as individuals approach retirement some level of de-risking of the investment policy is appropriate.
However risk is not simply a function of the growth asset exposure in the portfolio. Path dependency of returns is extremely important in the latter stages of pre-retirement and the early stages of post-retirement. In essence, pre and post retirees aim to maximise real returns subject to a low level of absolute volatility. This is not necessarily a function of the asset class exposure. In fact right now as we are likely in a world characterised by a global government bond bubble (ex-Australia), suggesting that the low risk portfolio is one dominated by government bonds is difficult to comprehend.
Life Cycle funds, which start with a high level of growth asset exposure that steadily reduces as retirement approaches (usually at a pre-determined rate) are not a good proxy for the changing risk tolerance and return requirements for accumulation members.
The historical record of Life Cycle funds for an Australian accumulation member is substantially worse than that of a fixed asset allocation through time.
A more intuitive approach would be to bring the investment policy back to the original objectives – maximising real returns with the minimum level of absolute volatility. Static (or pre-determined) asset allocation policies which have no reference to short to medium term expected risk and return are unlikely to be an appropriate accumulation or decumulation option for individual members.
The preferred model should be is for those approaching retirement to transition into a range of outcome orientated investment strategies (e.g. CPI+2, 3, 4, 5 etc) with appropriate downside risk metrics rather than fixed asset allocation strategies (including Life Cycle and Target Date funds).
Specifically a move to Lifecycle funds as considered in MySuper or more generally by the industry as has occurred in the United States would not appear to be in members’ long term interests. While such strategies sound intuitively appealing they don’t stack up in practice. Good marketing spin, poor investment underpinnings.
1Samuelson, Paul A., “Lifetime Portfolio Selection by Dynamic Stochastic Programming,” Review of Economics and Statistics 51:239-46, 1969, reprinted in Joseph Stiglitz, editor, The Collected Scientific Papers of Paul A. Samuelson, Volume 3, MIT Press, pp. 883-90, 1972.
2Merton R.C. (1969) “Lifetime Portfolio Selection Under Uncertainty: The Continuous Time Case”, Review of Economics and Statistics, 51, 247-257
3Anup K Basu and Michael E Drew, “Portfolio Size Effect in Retirement Accounts: What Does It Imply for Lifecycle Asset Allocation Funds?”, Journal of Portfolio Management, Spring 2009.
4Shiller, Robert. “Lifecycle Portfolios as Government Policy.” The Economists’ Voice, Vol. 2, Article 14 (2005).
5John Spitzer and Sandeep Singh, “Shortfall risk of target-date funds during retirement.” Financial Services Review 17 (2008). pp 143-153.
Opinions, estimates and projections in this article constitute the current judgement of the author as of the date of this article. They do not necessarily reflect the opinions of Schroder Investment Management Australia Limited, ABN 22 000 443 274, AFS Licence 226473 ("Schroders") or any member of the Schroders Group and are subject to change without notice. In preparing this document, we have relied upon and assumed, without independent verification, the accuracy and completeness of all information available from public sources or which was otherwise reviewed by us. Schroders does not give any warranty as to the accuracy, reliability or completeness of information which is contained in this article. Except insofar as liability under any statute cannot be excluded, Schroders and its directors, employees, consultants or any company in the Schroders Group do not accept any liability (whether arising in contract, in tort or negligence or otherwise) for any error or omission in this article or for any resulting loss or damage (whether direct, indirect, consequential or otherwise) suffered by the recipient of this article or any other person. This document does not contain, and should not be relied on as containing any investment, accounting, legal or tax advice. Schroders may record and monitor telephone calls for security, training and compliance purposes.