Understanding the journey to retirement
In our recent paper “Why strategic asset allocation is flawed”1 , we questioned the ability of the fixed asset allocation approach common to our industry to deliver consistent investment outcomes in line with objectives typically outlined to individuals. Subsequently2 we also showed that in order to meet individuals’ investment objectives a significant degree of flexibility in asset allocation was required.
One aspect of long term saving that is often overlooked is the need to focus on consistency of returns through time and the impact this can have on an individuals’ accumulation or decumulation payments. In this paper we show that due to the pattern of saving and dis-saving, achieving return objectives over very long term time horizons is simply not enough to meet that individual’s investment requirements.
Individuals receive money weighted returns not time weighted returns. This necessitates controlling medium term volatility of investment outcomes to a much greater degree than is recognised by the industry. The fixed strategic asset allocation process common in our industry contributes to this volatility and so runs the risk of delivering poorer investment outcomes. If we are to meet individual investment objectives consistently for different age groups and savings patterns through time a greater focus on delivering more consistent medium term investment returns is required. This will necessitate much greater flexibility in asset allocation relative to a fixed asset allocation approach.
As a way of introducing the concept of path dependency, let’s consider the hypothetical examples of Ray, Mardi and Chris using actual historical data. All had considerable foresight to decide that saving for their eventual retirement was a prudent course of action.
Ray was born just prior to the turn of the century and started saving in 1919 at the rate of 12% of salary. By the time Ray retired in 1958 after 40 years of contributions, his retirement nest egg had grown to 9.1 times his final salary at that date (his salary increased every year at inflation plus 2%). The average real rate of return earned by his fund over this period was 5.92% p.a.
Mardi on the other hand was born 40 years later, started saving just as Ray retired, and recognising Ray’s great savings outcome adopted the same approach of 12% of salary contributions (Mardi’s salary also increased at inflation plus 2% pa.). By the time Mardi retired in 1997 after 40 years of contributions her retirement benefit had grown to 13.6 times salary – an outcome nearly 50% better than Ray’s. However, her average real rate of return over this period was almost identical to Ray’s at 5.99% p.a. Chris starts saving in 1935 on the very same savings and investment strategy. Unfortunately Chris’s average real return over the ensuing 40 year period (from the same strategy) is only 2.05% p.a. and his retirement benefit has grown to only 4.8 times his final salary – less than half of Ray’s outcome. In this case even a 40 year time horizon wasn’t sufficient to generate anywhere near the expected outcome (which at CPI+5% would be closer to 10 times salary).
Chart 1 below shows for all three individuals the year they commenced saving, the average return over the period and the resulting end accumulation as a multiple of their final salary.
So why the difference in outcomes when average returns were so similar? In the early part of Ray’s savings period real returns were very strong, while in the latter part, the 1940’s and the 1950’s, returns were somewhat weaker. In contrast, Mardi’s final years of employment occurred in the 1980’s and 1990’s, a period of very strong real returns. Given that the account balances of Ray and Mardi are at the largest in the later years of their careers it is these periods that have the most impact on final outcomes. Chris clearly suffers from the issue of just being born at the wrong time.
Almost identical average real returns over 40 years (for Ray and Mardi), the same savings pattern, and very different results. Clearly the investment journey taken by Ray and Mardi has had a major impact on the outcome. While in this case, partly because average returns have been so strong (6% p.a. real is well above the long term average), the results in both cases are quite good, the dispersion of outcomes for the same strategy is not. Likewise for Chris, if the year of birth can also have such a significant impact on the final result how can the industry confidently outline to individuals the likely results of their savings approach?
In prior papers we considered the extent to which the fixed strategic asset allocation approach, which is the mainstay of the investment portfolio construction of most investors, was able to meet the investment objectives and to understand the real risks introduced from such an approach.
We showed that over the last 110 years a traditional 60/40 balanced type portfolio has generated real returns of around 5.3% p.a. However:
- the medium term volatility of the results was significant, with the best 5 decades accounting for an average real return of 9.1% p.a. and the remaining 6 decades giving an average real return of only 1.5% p.a.;
- there were substantial periods of drawdown relative to inflation with one such period lasting as long as 12 years and resulting in a loss of value relative to inflation of over 45%; and
- substantial flexibility is required in asset allocation in order to dampen the volatility sufficiently to achieve individual investment objectives consistently (or at least over rolling 5 to 10 year time frames).
The medium term volatility of the fixed strategic asset allocation based approach is highlighted in Chart 2.
It is more than just time horizon
The argument commonly adopted by the industry is that despite the objective of generating more consistent medium term outcomes, in reality these are not that important. What is important is that the return generation in the very long term is sufficient to meet member expectations. The example we used at the start of this paper highlighted that similar average results over time can still lead to very different outcomes for individuals – premised entirely on the luck of the draw as to what year they were born.
We noted in our SAA paper the time horizon required to say with 90% confidence that we would achieve a target return of CPI+4% was 53 years and CPI+5% was 85 years. The best we could hope for with a 40 year time horizon and 90% confidence was only CPI +2.6%.
However, even if did have lower expectations and a very long time horizon is it really the case that we could ignore these medium (10 year) time frames? We disagree with this “very long term view” for a number of reasons:
- “The long term” is far longer than most investors’ time horizons. Based on our historical balanced portfolio, even with a 20 year time horizon, there is only a 90% probability of delivering CPI+2.2%. Achieving CPI+5% with a 90% probability required a time horizon of 85 years (almost the entire sample period of 110 years).
- There is considerable behavioural finance literature that suggests that shorter term time frames are important for individuals, if only to stop potentially poor decision making that may see individuals alter their investment strategy after periods of poor performance and so capitalise those performance losses. The very sharp uptake in annuitisation products that we observe today is, in our view, part explained by investors experience of recent poor market performance. This is potentially adverse when we consider that long bond rates have collapsed (on which annuity rates are based) and risk assets offer greater prospective returns (relative to say 2007). Consequently locking in a “guaranteed” low rate at this point is not necessarily wealth maximising behaviour. Agnew et al3 examined the take-up of annuitisation products in an experimental framework after simulated periods of good performance and poor performance from equity markets and found historical performance experience had a significant impact on future investment choice.
- The accumulation and decumulation process for individuals means that there is a constant application of new capital (or the reduction in capital) through time. This process means that individuals do not earn long term time-weighted average rates of return that are described by the industry. Individuals earn a money weighted return which will be more heavily influenced by performance in those periods of time when the capital value is at its highest.
- As with individuals, comingled funds (such as superannuation funds) do not actually earn the long term average rates of return they advertise either. Again these are time weighted returns and, except in the unique case where total cash-flow to or from a fund is zero, the money weighted return actually earned by the fund will be different from the time weighted return shown. Consequently, the purpose of this paper is to examine the degree historically to which money weighted returns earned by individuals may differ from time weighted returns. In particular we look at the historical difference in end investment outcomes to individuals that can result from the medium term volatility of performance combined with an accumulation or decumulation process.
It is as much about the journey
The typical response of the industry to this medium term volatility is that most individuals, particularly those saving for say retirement, have a time-horizon that is much longer than 10 years and consequently as the time horizon is extended this volatility of returns is reduced. However, recognising that the typical pattern of saving and dis-saving is for gradual contributions or drawdowns to be made throughout the period, this has a substantial impact on the real time horizon of the individual. In particular, the “journey” of returns earned over the savings (or dis-savings) period can have a substantial impact on the end result even where the average return may be in line with objectives.
Time Weighted vs Money Weighted Returns
The potential difference between time and money weighted returns can be best explained by example. Consider the simple example in Table 1 of two investors, Lisa and Darren, who make contributions of $100 per year and earn an average 7% p.a. over a 3 year period. While the average returns are the same, the pattern of delivery of this return is very different as shown in the following table.
We can see in Table 1 that while the average time weighted return for both investors is 7% p.a., for a given total capital contribution of $300, Lisa has earned $43.99 in return over the period while Darren has actually lost money relative to his total capital contribution. A fund that earned these returns however would have presented the return in both cases as 7% p.a. Interestingly the arithmetic (simple) average return for Lisa is 7% but for Darren is 10%. So despite Darren having a higher simple average, his returns are still lower. What we observe in Table 1 above is the impact volatility of investment earnings can have on the final accumulation when there is a process of capital contributions being made (i.e. what occurs in most real life cases).
To further illustrate this point of the weight of capital on the end accumulation, Chart 3 shows the effect of contributions and investment returns in the first 20 years and second 20 years of a member’s working life. In the first 20 years a 1% increase in the annual contribution rate will have broadly the same effect as a 1% increase in the annual investment return. However, in the last 20 years the situation is considerably different.
Impact of the accumulation process on money and time weighted returns
Recognising the constant contribution based nature of the accumulation process (e.g. 9% or 12% of salary per year), one of the potential issues faced by any savings vehicle that utilises a fixed asset allocation across its membership base is the impact that the accumulation process itself will have on the end value. In particular, to what extent will the volatility of returns impact an individual’s end accumulation versus an individual earning the same (geometric) average return over that period but with no volatility?
In both cases we assume the member makes the same contribution as a percentage of salary and salary increases at the same rate each year. Chart 4 highlights how on a rolling 20 year basis, the year by year accumulation (money weighted) can result in substantially different outcomes to an individual who earned the same constant average rate for that 20 year period (termed “time weighted”).
Clearly there have been some quite differences between the average rates of return from a strategic asset allocation approach and the eventual earning rate for the individual given the accumulation of capital over time. We also note that in Chart 4 those commencing accumulations in the 1980's and early 1990's will have generally experienced a worse money weighted return than is being suggested by the industry.
In any case, bearing in mind that it is really the difference in results that matters than which does better or worse, we have shown below in Chart 5 the range of outcomesover 10 and 20 years. It is clear from the chart that as the time period is extended the potential volatility of year by year earning rates can result in very meaningful differences in the end outcome to individuals.
Impact of decumulation process
In the same way that the volatility of year by year earning rates in the accumulation process can have a significant impact on final accumulated value, we can also examine the impact of volatility on members in the drawdown process. This is potentially more serious for the individual as for many savers in the accumulation phase there at least exists the opportunity to alter either the saved amount or extend the duration of saving and so skew the potential outcomes to a level where the individual is more comfortable. However, for individuals in the drawdown phase, often this flexibility does not exist. While drawdown amounts could be changed throughout the drawdown period, this in part negates the objective of ensuring a certainty of drawdown amounts and may have significant negative lifestyle consequences.
In Chart 6 we show for an individual commencing a drawdown in any year from 1900 to 1990 where the drawdown amount is set at 7% of the initial capital and indexed with inflation, how long the individual’s capital will last.
We can see from the above that there is substantial volatility in the duration of drawdowns with the determining factor being the year the drawdown commences. Those individuals who commenced a drawdown in 1918 to 1925 would still be drawing down today. However those individuals who started the same drawdown pattern in 1968 to 1973 would have run out of money after 11 or 12 years. Even one year can make a major difference. An individual starting drawdowns in 1973 would see them last 12 years. The same process started in 1974 would last 35 years.
Put another way, historically there was a 35% chance drawdowns would have lasted less than 20 years and 30% chance they would have lasted more than 30 years. That is a wide dispersion in drawdown results for the same starting conditions. It is unlikely that presenting these potential outcomes to a retiree would instil confidence in the chosen investment strategy!
The above highlights the need to be mindful of medium term volatility of returns as the impact on individual investment outcomes can be significant.
Consequently, it is very important to minimise downside volatility for the older demographic. While a simple model is to reduce the exposure of the older demographic to risky assets, the episodic nature of equity market returns makes this problematic - with periods of high returns followed by periods of low returns. If the period of high equity market returns is at the end of a career, this model of reducing their exposure to this asset class toward the end of their career has the potential to severely limit the individual’s ability to accumulate a lump sum for retirement. We considered this further in an earlier paper “Life Cycle Funds - Just Marketing Spin”4.
More generally the fact that asset class risks are not stable through time makes any approach to reducing risk that is premised on a fixed relationship between asset class exposures flawed. Rather, the emphasis should be on reducing the incidence of downside volatility not necessarily a particular pre-determined asset allocation that bears no relationship to forward looking returns.
This would suggest that for plans dominated by an ageing demographic structure, downside risk management is very important. However, it could be argued that this should be the status quo no matter what the demographic breakdown of a plan. Given members of the various demographic cohorts will generally be present tilting the fixed asset allocation to suit the dominant cohort will raise intergenerational transfer issues.
Managing to a return objective and not to a fixed asset allocation will help to minimise the impact of the episodic nature of returns from equity markets and therefore the fixed asset allocation model. Importantly this would reduce the impact the luck of the year of birth has had in historical results.
The purpose of this paper was to investigate the additional volatility introduced into individual investment outcomes as a result of the accumulation or decumulation process.
While over the very long term a fixed strategic asset allocation process can deliver satisfactory real return outcomes, few investors actually ever receive these “average” results. In particular, the substantial volatility of medium term return outcomes that a fixed asset allocation process delivers means that the potential volatility of time weighted average outcomes is large.
Fixed asset allocation portfolios require a very long term time horizon, given that equity markets in particular have delivered real returns in long term cycles or ‘regimes’. This has important implications for the money weighted returns earned by individuals. Importantly, the process of accumulation and decumulation means that individuals rarely have constant capital throughout their investing time-horizon and this process results in additional volatility of potential return outcomes.
We note that the accumulation nature of savings for retirement sees market movements toward the end of the accumulation stage having an outsized impact relative to earlier years in the accumulation stage. This sees limiting the downside volatility as a key requirement for older demographics. Skewing the fixed asset allocation of an approach to benefit any demographic, even if it is the dominant one, raises intergenerational issues. The solution to these issues is to manage to a medium term consistency objective, which is in line with our conclusions more generally on the impact of fixed strategic asset allocations.
Once again, if the objective of the industry is to improve the consistency of individual investment outcomes, substantial flexibility is required in the asset allocation process.
Opinions, estimates and projections in this report 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 ("SIMAL") 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.
SIMAL 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) or 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. Past performance is not a reliable indicator of future performance. Unless otherwise stated the source for all graphs and tables contained in this document is SIMAL. For security purposes telephone calls may be taped.
1“Why strategic asset allocation is flawed”; Cooper, Durack, Doyle and Stevenson, Schroder Investment Management Australia Limited, March 2012
2“Asset Allocation: How flexible do we need to be?”; Cooper, Durack, Doyle and Stevenson, Schroder Investment Management Australia Limited, April 2012
3“An Experimental Study of the Effect of Prior Market Experience on Annuitization and Equity Allocations”, Feb 2010, Agnew, Anderson and Szykman.
4“Life Cycle Funds - Just Marketing Spin”, Greg Cooper, Schroder Investment Management Australia Limited, September 2011
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.