It's about risk not return
In times of significant market stress it is always tempting to look for the panacea and critique what hasn’t (or doesn’t appear to have) worked – and flock to what has. The sharp downturn in listed equity and credit markets has left many investors questioning their ability to take risk.
The last 18 months has been a period of “the great unravelling” in financial markets. Globally the appetite for risk has contracted sharply. Investors who were once happy to gear into equities, buy complex structured credit derivatives and even those with simple “set and forget” risk dominated investment strategies have been flocking to the safety of cash (especially government guaranteed cash). Strategies once considered to be “diversified” have been shown to be highly correlated. Many long term investors (such as superannuation funds) have suddenly needed to focus on short-term risks – particularly liquidity – as calls on commitments to unlisted asset investments and losses on currency hedges have soaked up more than the positive cash flows from members and realisations on “cash” investments could provide.
We would agree that in times of extreme volatility cool heads should prevail. However, we should also ask ourselves if as an industry we properly manage for risk? Are we appropriately devolving our responsibility to investors to manage the risks of the investments with which they have entrusted us? Already 3, 5 and 7 year returns on an upper-quartile superannuation fund is sub-cash.
Fundamentally the risk of a superannuation fund (including their Trustee boards and their investment committees) comes back to two broad principals:
- Can we pay our investors back their money when they want it; and
- Will we have added any “real” value to that money while we had it?
The implications for our industry of one or both of these not being possible for even a short period (a few weeks in the case of the first and say 2-3 years in the case of the second) are quite severe.
At the total fund level risk is generally described and reflected in a funds’ Strategic Asset Allocation (SAA). This SAA is often fixed (or largely fixed) for a long period of time. Our industry largely ascribes the riskiness of a fund as represented by its SAA – in particular its allocation to “equity” type growth assets and “bond” like defensive assets. The entire “balanced”, “growth”, “conservative” framework on which many investment options are classified is predicated on fixed exposures to different asset types.
However, in our view, while the SAA of a fund may be static, the risk of that SAA is anything but static. The typical response to this is to suggest that some form of benchmark is needed to classify risk and over the long term the outcomes represented by different asset allocations have generally been reflected in different returns (the higher risk asset allocations generating higher returns).
The problem is that risk is often described ex-post (e.g. tracking errors, standard deviation, sharpe ratios etc), but management of risk is required ex-ante. Ex-ante risk management at a total fund level requires a much greater focus on qualitative analysis – understanding how an asset has changed its risk characteristics either by the actions of the managers of those assets (e.g. listed property trusts increasing gearing) or the actions of other investors in those assets (the “crowded” trades). Risk management at this level has to focus on the real drivers of risk – operating (including economic) leverage, financial leverage, liquidity risk and valuations.
Natural human behaviour is to seek out success and reward it. Witness fund flows into the better performing (in absolute terms) products over time or the “God-like” status with which we revere the biggest risk takers in business (well the ones who were successful). Our brains and hearts are avid buyers of selection bias. See the hedge fund industry as exhibit A. Enough said.
Long-term successful investing is much more about managing risk than seeking returns. Return estimation over shorter time horizons (especially the time horizons most in the industry look at to measure success) is extremely difficult. The standard error in the forecasts far exceeds the forecast itself. However, the secret to long term investment success is steadily compounding positive returns over time. In particular it is about avoiding losses, especially catastrophic losses. A 50% drawdown requires a 100% return just to get back to square. As such, the focus on investment policy should be on risk. Get risk right and let the returns look after themselves.
Unfortunately our industry concentrates on accentuating the outcomes – particularly return outcomes. We make constant reference to the performance league tables – tables that usually don’t include any realistic measure of risk (historical or prospective).
If the last 18 months has taught us anything it should be that real risk management of portfolios is critical to achieving good long term performance outcomes. The greatest driver of risk in portfolios will be the underlying asset allocation. As such, to properly manage risk it is necessary to take a more dynamic approach to the management of a funds’ asset allocation over time. Unfortunately the implementation and governance structure around most funds today makes the implementation of real risk management somewhat difficult. As an industry we have spent 20 years devaluing asset allocation skills and unwinding the structures that made dynamic asset allocation possible. It is time to change that.
At the total fund level, the embodiment of the risk profile is typically the Strategic Asset Allocation.
Our experience in Australia is that indeed most funds have adopted a relatively static SAA policy, and despite regular reviews this has largely remained unchanged through time. Indeed most PDS’s outline their risk and return expectations to members/investors in the context of a fixed long term SAA.
The traditional approach to setting an investment strategy is that the SAA is set by making a series of long term assumptions about asset class returns and volatilities, then optimising this to determine the “best” portfolio to achieve the investment objectives. While in principal such an approach is intuitively sensible it unfortunately ignores the fact that the “long term” in most SAA assumption setting environments is in fact the very, very long term (probably 30+ years). Over a more “normal” period, as risk premia in markets change, expected returns can vary significantly from these long term assumptions and risk – meaning the likely range of returns around the expected outcome – can be significantly skewed upwards or downwards.
In these periods when risk premia change, the required outcome – especially in the case of a portfolio aiming to deliver positive real returns through time – is unlikely to have moved to any great degree. This implies that under a typical SAA approach the likelihood of delivering the real return outcome would shift materially as risk premia adjust.
On this basis, it no longer sounds intuitively sensible to describe a fund’s risk and return expectations in the context of its SAA. In our view, if a change in risk premia on an asset means that it is significantly more likely to achieve the required return outcome relative to other assets then its weight in the portfolio should change materially.
As such, the starting point for portfolio construction should be significantly more dynamic than that in traditional portfolios, and is particularly sensitive to the risk premia embedded in asset exposures at a particular point in time. In addition, as risk premia change, the asset allocation should potentially change (in some cases meaningfully).
Unfortunately, this approach flies in the face of the generally accepted advice most funds have received over the years, and requires a completely different “structure” to the way most funds are currently structured. The notion of broadly fixed asset allocations to range of different “asset classes” managed by underlying asset class specialists doesn’t make sense when we have the potential for a significant change in asset allocation (or it is certainly more difficult to manage). In particular, funds still spend significantly more effort on selecting the specialist to manage an asset class than they do on whether it is the right asset class to own (or keep owning).
Of even more importance (for trustees and investment committees) is how to make the decisions on asset allocation in this framework – the process for which has been gradually taken back from asset managers over the last 20 years and placed firmly in the hands of the trustees and their advisors.
The notion that a static SAA represents a constant level of risk (and return expectation) is, in our view, a flawed one.
Strategic Asset Allocation and the variability of returns
The concept of the long-term investment strategy as the embodiment of a funds risk profile can be broadly traced back to the benchmark Brinson Beebower and Hood (BBH) study printed in the Financial Analysts Journal, July/August 1986.
The BBH study concluded:
“…the investment policy return…explained on average fully 93.6 per cent of the total variation in actual plan return; in particular plans it explained no less than 75.5 per cent and up to 98.6 per cent of total return variation. Returns due to policy and timing added modestly to the explained variance (95.3 per cent), as did policy and security selection (97.8 per cent)…..total return to a plan is dominated by investment policy decisions. Active management, while important; describes far less of a plan's returns than investment policy.”
This six page article has been used (and mis-used) to present a variety of different views on long-term portfolio structure. This study, and the many updates or derivatives of it, form the basis for the notion that all funds should have a long term investment strategy – defined as the SAA – and that while the SAA is subject to regular review it should not vary significantly through time.
The principal finding of the BBH study (and most since) was in fact that the asset allocation is the principal determinant of the variability of returns over time. To this we would agree entirely. A fund’s broad asset allocation will largely determine the variability of returns – stock selection and minor changes to asset allocation around the broad asset allocation will, except in extreme cases, have little influence on the variability of returns through time (e.g. even those managers who outperformed in Global equities in October would have done little to influence the overall return for a market that was down 20%).
However, this is not to say that the SAA should remain static through time. In fact if the broad asset allocation is likely to describe most of the return variability (and by extension, risk) of a fund then we would argue that the management of the asset allocation is by far the principal consideration that trustees (including investment committees and their advisors) should have.
In particular, where the medium term expected return (and risks) can vary significantly, consideration as to the appropriate asset allocation in light of a funds risk and return objectives should be the primary focus with respect to investment decision making. While this process has largely been “outsourced” via the establishment of an SAA, we would argue that this absolutely does not absolve trustees from their obligations to members to manage risk.
The evidence - skill or luck?
History is replete with examples of those who have been branded heretics for deriding the popular wisdom of the time. And so it is that by deriding the concept of a fixed SAA, particularly as a way to measure and control risk and a concept we have been drilled into accepting over the last 20 years, we are indeed in danger of being branded the same.
Galileo was ordered imprisoned (subsequently commuted to house arrest) for his views on heliocentrism (that the Earth revolved around the Sun) rather than the geocentric view that had persisted since the time of Aristotle. While we are not proposing an end to a philosophy with ramifications as far reaching as the movement of heavenly bodies, one suspects getting the industry to reconsider its position will be a herculean task.
The traditional defence for having an SAA is that the alternative (which may be labelled market timing) is difficult and the evidence is such that it doesn’t work. To the first point we would agree, however the extent to which it is difficult is more a function of the time horizon over which you are aiming to be correct.
This leads to the second point – we would agree that much of the evidence on practices labelled as “market timing” is not particularly convincing. However once again it is important to understand the objective and time horizon against the outcomes.
Let’s divert for a moment and consider the catch-cry most commonly rolled out to investors in times of significant market (downward) volatility.
It’s time in the market, not market timing
It’s a nice marketing phrase, and one often used by the retail facing part of the industry. Unfortunately it’s also wrong.
Take the common presentation of this statement, which is usually along the lines of “if you had missed the top 10 days of market performance over the long term you’d be xx% worse off” – or the various variants on this. That statement is indeed in itself correct. As the charts below show for Australia and the US, if you had missed the top 10 performing days in the last 28 years for Australia or 45 years for the US – you would indeed be worse off, and substantially so.
In the Australian example, an investor missing those top 10 trading days would have been 40% worse off over the accumulated period and our US investor over 48% worse off.
Unfortunately, much like the set and forget logic of SAA, the logic of the above analysis is fundamentally flawed. Firstly it assumes that our hapless investor who misses the top 10 trading days would have been fully invested in the lead up to those days, and fully invested thereafter (as well as no trading costs etc – but lets call that noise for the sake of simplicity).
While there are some investors who may have managed this feat of market timing, I suspect that even for a “professional” this would be somewhat difficult to pull off. Interestingly, if we look at when these top 10 days occurred, those with an eye for detail may notice a somewhat disturbing relationship.
Top 10 Performance Days for the Australian and US Equity Markets
|Australia – All Ordinaries||US – S&P 500|
Source: Schroders, Thompson Datastream, through to 31 October 2008.
Most of the key dates above fell in periods when overall market returns were quite poor. We can see the significance of the above by reconstructing our “missed days” analysis above and assuming that not only did we miss the top 10 days in the market over the period but lets assume we missed the 100 prior days and the 100 days thereafter (i.e. Got out early and didn’t invest until well after the event).
All of a sudden the notion that it’s “time in the market” starts to take on a different twist. Now, by missing the 7 odd months around each of these significant market “upticks” our returns have improved significantly. Our hapless Australian investor who missed the top 100 days either side of the best day in the market is now 144% ahead of the index (and that is not even allowing for a cash return on their money while out of the market) and the US investor 73% ahead.
Is this just data-mining or is there something more to it? Looking at some other scenarios, if you just missed the year in which these “top performing” days occurred, you’d be 3% better off in Australia and 17% better off in the US. Too long a time frame? Well, if you missed the 5 days either side you’d be slightly better than 20% off in both markets over these time frames.
And all of that to say nothing of the volatility reduction (which is significant) from missing these spikes. “Time in the market” doesn’t look like quite the no-brainer we make it out to be anymore.
Strategic Asset Allocation – rest in peace
The foundation of a plan’s SAA is often treated as the defining moment for the encapsulation of a plan’s risk and return objectives. The process behind the establishment of the SAA involves much thought about performance objectives, risk tolerance and (often) the conducting of an Asset and Liability Modelling (ALM) study. Charts are produced, efficient frontiers are discussed and ultimately a single asset allocation is selected as the embodiment of the long term allocation to various asset “classes”. While this process may be carried out regularly every 1 to 3 years, in reality updated SAA positions vary little from year to year, and often only vary to the extent that new asset classes are added. The problems with the typical approach to SAA are numerous.
It could be argued that we are raising the concept of questioning the SAA only after a period where it appears not to have worked. It is at these times that our industry rolls out the charts showing the long term growth in equities and how it only took 2 years for markets to recover the ground they lost in 1987 (the last sharp downturn in equities most of us in Australia can remember). To the end of December 2008, 5 to 7 year returns on an upper quartile “balanced” superannuation funds is circa 5-6% p.a. and 10 and 15 year returns are similar (again 5-6.5% p.a.) – not much (and in the short term) less than cash rates. i.e. with all the complexity, fees, and intellectual firepower aimed at the superannuation industry, many of our investors would be better off having left their money in the bank.
While we can argue this is a point in time statement and (hopefully) risk assets have seen their worst performance behind them, over most other time horizons the strategic positioning of most funds looks to have been broadly right – it has delivered what it set out to deliver. If it ain’t broke, don’t fix it. That said, it is pertinent for us to ask the question most commonly asked of the active managers most of us employ – was that luck or skill?
To more objectively answer that question, we can break our analysis down as follows:
- What assumptions are we making?
- What happens if we are wrong?
The typical SAA approach involves the setting of a base level of assumptions for asset market return and risk expectations (and the interaction between asset classes – or correlation). These are then typically analysed in a mean-variance framework to determine an appropriate portfolio that meets the risk and return objectives of the fund. While this sounds (somewhat) reasonable, the first problem is that the assumptions don’t really seem to change that much. Expectations for equities vs bonds vs cash tend to follow a relatively consistent relationship. We almost always assume equities will beat bonds and bonds will beat cash. But they don’t, and they don’t for very long periods of time.
The 10 year return to date on US equities is negative. And it was negative before we got to October 2008. In fact with the exception of Australia, a large number of equity markets are now showing 10 year negative returns.
More basically we should even consider the notion of an asset class. What in the first place makes an “asset” special enough to be considered a “class”? Over the years we have seen the adoption of numerous new “asset classes” in portfolios – infrastructure, private equity, hedge funds, absolute return strategies, commodities – to name but a few. Inevitably at the point we define something an asset class, it ends up in the SAA.
This also appears to be somewhat independent of price. Once an asset has been accepted as part of the SAA the price at which subsequent investments are made is independent of the price. If you want to know which “asset class” is primed for a substantial period of outperformance – find out when it’s being defined as an “asset class”.
While the assumption setting process is difficult, it is not impossible – provided time horizons (and expectations) are set appropriately. However it is important in the process to recognise that the biggest driver of shorter term equity market returns is likely to be valuations (e.g. PE ratios) and consequently as markets rise future expectations for market returns will fall. Consequently the assumption setting process needs to be dynamic in the sense that expected market returns can and will move significantly through time. They are not static.
Does it matter?
While it would be nice to assume that even if we are wrong (and indeed as far as assumptions are concerned we are almost definitely going to be), does it actually all matter? Over the last 100 years the return to a typical balanced type portfolio has broadly delivered its real return objective.
The last few decades have been without exception in the last 100 years the longest running, most profitable environment for investors with a large exposure to “growth” type assets (e.g. equities, private equity, property etc).
The average real return on a typical balanced fund for the 1980’s, 1990’s and (minus the last few months) in the 2000’s has been in excess of 8% p.a. – well above the long term average.
However, we have to be conscious that this “boom” time for equity (and indeed all financial) assets is, while slightly unusual in its magnitude and duration, not without historical precedent.
The 1950’s and the 1960’s (the “post war boom”) saw similar levels of returns to a typical balanced portfolio (to the extent they existed). As with the last 20 years, this was also a period characterised by steadily declining inflation, relatively strong and stable GDP growth and (not surprisingly) general economic optimism.
Unfortunately we then hit the 1970’s which were characterised by sharply rising inflation, falling (albeit volatile) global growth leading to very poor investment returns. A typical balanced fund in the 1970’s would have produced a decade long real return of -2.8% p.a.. I wonder how many in our industry would survive that type of return? How would we justify to investors to “take a long term view” when they’d had 10 years of -2.8%p.a. real?
The question we should ask ourselves as longer term investors is indeed, is it different this time? Can we continue to extrapolate the boom economic conditions of the last 25 years into the future, or should we expect economic conditions to fluctuate as they have done consistently in the past?
The chart below might help somewhat in putting that question into context.
It is quite clear from the above that the prevailing economic environment is not without precedent, but also that the environment is prone to change (and change quite dramatically). Investors who believe that inflation is likely to remain low and stable and growth strong and stable may be right, but it appears the odds are stacked against them.
We believe that the disinflationary tailwinds that have strongly supported financial assets – particularly equities – over the last 2 decades are behind us and this will impact asset markets going forward. While we could argue that the repricing of listed equities in the last 12 months has made them cheap – that “cheapness” could persist for a long period of time (as it has done previously).
To get a sense of the possible implications from a portfolio perspective of this we have shown below the decade long returns and real returns of a typical 70% equity/30% bond portfolio over the last 100 years.
|Decade||Australian inflation,% pa||Return, % pa||Real Return, % pa|
|Last 100 Yr Avg||4.0||9.2||5.4|
Source: Schroders, Datastream
What is clear from the table above is:
- Longer term real returns from a strategic “70/30” portfolio have been in the order of 5-6% p.a. – justifying the broad objective of most superannuation funds today.
- However, there have been periods (generally corresponding with a change in the prevailing macro-economic environment) where the strategic return has been substantially negative or very low for a long period of time.
So, getting it wrong matters, a lot.
What way forward?
So where to from here? Having established that a fixed SAA and the framework around which we build most portfolios is certainly in need of review, we would propose a few initial steps that can and should be taken:
- Start to re-introduce asset allocation or consideration as to asset allocation views at the trustee/investment committee level. I would surmise that in many cases this is not going to be in the remit of existing advisor relationships.
- Establish a framework around which asset allocation decisions can be made. The first step in this will be appropriately setting the time horizon for success (this is not a Tactical Asset Allocation (TAA) marketing story, the time horizon needs to be realistic – say 3-5 years). Secondly, establish either the process for making such decisions and appoint internal or external expertise to assist in this.
- Examine and review total portfolio risk at an aggregate level based on real forward looking return and risk assumptions – utilising market information on returns that is available today (e.g. current risk premia in assets, not 30 year return expectations). Compare this with what has been communicated to investors.
At a fundamental level we need to reintroduce skills that have been largely lost to the industry over the last 20 years – real medium term asset allocation decision making skills.
Strategic Asset Allocation has been the bedrock on which the superannuation/pension/investment industry has established itself over the last 20 years. The SAA typically represents the embodiment of a fund’s risk and return profile.
However, risk premia in markets change – and substantially so through time, whereas SAA’s typically remain quite static. As investors, our job is to manage and weigh risks in portfolios, a fixed SAA framework does not allow for the appropriate weighing of risks.
While SAA’s in Australia appear to have worked over the last few decades in terms of the outcomes delivered we should consider that this was through an exceptionally positive environment for risk assets (particularly equities). The creation of a long term portfolio, and the understanding of the risks in that portfolio, needs to specifically take into account the likely headwinds facing financial markets now more so than we have seen over the last 2 decades.
While our “clients” may tolerate one, two or even three years of negative or sub-CPI performance, one questions the behavioural implications from a significant period of underperformance in equity markets. On the positive side, recent market movements have seen a steady re-pricing of credit risk, the resumption of illiquidity premiums and, ex-Australia, the re-introduction of an interest rate term premium. Equity markets have begun to respond to these inputs and re-price downwards.
These changes in asset prices make the achievement of long term real return objectives achievable over the medium term, and without a significant shift out the risk curve. However, what is required is a shift in thinking from investors that the process of a fixed strategic allocation to a number of equity like asset classes is the most appropriate way to achieve the performance objective.
Albert Einstein defined insanity as doing the same things over and over and expecting different results. Will the standard approach to building portfolios satisfy real return objectives if the market environment (especially for equities) remains weak? We believe a static approach is unlikely to deliver the required results going forward. While a more dynamic approach may not be market convention (and indeed sounds like a return to the days of balanced portfolios) different times call for different solutions.
Now is one of those times.
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.