Does strategic asset allocation make sense?

Strategic asset allocation may represent the conventional approach but Dominic McCormick questions whether it deserves the credibility it carries.

At the recent PortfolioConstruction Conference a survey of the audience showed that 57 per cent used strategic asset allocation (SAA) as their asset allocation approach for clients.

Ironically, the conference contained much discussion about the need for more flexible/dynamic asset allocation (DAA), especially in the current challenging investment environment. 

However, this article is not really about the case for more active asset allocation even though it is what I have believed in, previously written about and practiced for most of my years in the investment industry (after overcoming the brainwashing of my early economics and finance studies).

Nor do I suggest that successful active asset allocation is easy.

Rather, this article is a critique of the SAA approach which, as the survey suggests, retains the incumbent position for asset allocation approaches, but in my view, definitely does not deserve that role.

My argument does not rely on proving that DAA is better. Rather SAA should be abandoned, not because there are better approaches available, but because it is flawed to its core, always has been, and would never have gained credibility amongst investment industry participants if they had not been so easily seduced by the ‘efficient market’ view of the world in the 1980s and 1990s.

In fact, I challenge anyone to provide an acceptable, practically and theoretically consistent definition of what SAA actually is.

One that makes rational sense in the real investment world where assets become over and undervalued, bubbles and busts occur and where returns, volatilities and correlations vary markedly even over the long-term.

Often SAA is couched in vague terms referring to some ‘long-term’ asset allocation but with little further explanation.

Below are a range of definitions of SAA that are usually put forward, along with their inherent problems.

1. The ‘neutral’ asset allocation – neutral to what? By itself it is a meaningless concept.
2. The ‘normal’ asset allocation – normal compared to what? Another meaningless concept. 
3. The asset allocation that will meet an investor’s long-term objectives. This says nothing about how the SAA is determined, but in any case there is no predetermined ‘set and forget’ asset allocation that can be relied on to meet reasonable investment objectives. Crucially, the likelihood of any specific asset allocation meeting such set objectives will vary markedly over time.
4. The ‘constant risk’ allocation – if risk (whether volatility or more relevant measures such as chance of return below a certain level) is changing over time, then by implication, the appropriate asset allocation to maintain equal risk cannot be static. Only by actually changing asset allocation over time does one have any chance of keeping risk constant. Therefore SAA by definition cannot be constant risk.
5. The ‘efficient portfolio’ based on optimising return for a set level of risk using primarily historical returns, risks and correlations. This is the conventional way SAA has been constructed in the past but this backward looking approach is increasingly viewed as largely useless given returns, correlations and volatility vary markedly over time and valuations, not history, are one of the key drivers of future returns.
6. The appropriate asset allocation if all markets were valued fairly. This definition seems appealing at first until you realise that if just one asset class is overvalued or undervalued not only does this change that asset’s return forecast, but it also changes expectations for correlations and even volatility which would mean all other asset class weightings would need to change too. No support for SAA there.

How has a concept that doesn’t make rational sense become the key foundation for the way most financial planners (and institutional investors) implement asset allocation?

Part of the explanation is that SAA presents well and provides precision. Proponents can refer to a specific long-term SAA and illustrate the exact historical risk and return characteristics of that asset mix in nice charts.

It allows for precise performance attribution showing the value added or subtracted by any tactical asset allocation (TAA) around the SAA and that from investment selection within asset classes.

But this attractive analysis has been built around a deeply flawed concept which prefers being precisely wrong rather than approximately right.

In an uncertain world the desire to anchor to a long-term SAA for multi-year periods or even decades is understandable, but risk premiums for asset classes are not constant over time – even over the long-term – and are heavily dependent on starting valuations.

And that’s before even looking at correlations and volatilities – the other two inputs into most SAA models – which can also vary vastly over time.

It is simply ludicrous to think that the same ‘set and forget’ SAA mix that suited an investor last year or three years ago, suits an investor investing today or next year.

The only world where SAA makes sense is one where markets are efficient and always fairly priced and asset classes consistently deliver their ‘appropriate’ long-term risk premiums. If you still believe that is the sort of investment world we live in then you are highly vulnerable to all sorts of myths.

SAA was first developed by proponents of the efficient markets school and over-theoretical applications of Modern Portfolio Theory. The SAA religion was spread by asset consultants who, whether they believed in it or not, embraced it because it facilitated the performance analysis and other tools they could sell to clients.

The compromise for those who lacked belief in efficient markets was to allow for minor TAA moves anchored around the long-term SAA. But a flawed asset allocation approach is still flawed – even with minor allowance for the real world as an afterthought. 

Of course, most consultants and research houses still promote SAA to some extent, even though their faith in it is rapidly fading. After all, it is hard for them to openly admit that what they have been promoting to their clients for several decades is rubbish.

Instead they are quietly moving away from SAA by introducing ‘strategic tilting’, variable asset class risk premiums, limited ‘dynamic asset allocation’ or simply emphasising their TAA over/under weights more. This trend to de-emphasise SAA will continue in coming years. As usual, retail is lagging the institutional industry.

Some SAA supporters might counter with statements such as “I review SAAs often, even as frequently as quarterly or yearly and use forward looking analysis and not just history”. Fine, but such frequent SAA changes indicate you have already rejected traditional SAA and moved to flexible asset allocation.

Having said that, making asset allocation changes only on certain dates of the calendar rather than as needed in response to market valuations and other key drivers is far from optimal.

Ironically, I do still see some role for ‘static asset allocation’. Static asset allocation benchmarks or simple static diversified portfolios (systematically rebalanced) are useful to compare performance against over the long-term and/or to provide a ‘base case’ asset allocation approach.

Indeed, I can understand those who say they use a static approach because other approaches are too difficult to implement. However, a ‘static’ approach doesn’t pretend to have a robust theoretical or practical basis as SAA does and it doesn’t counter most of the concerns raised in this article.

Further, while clients might not expect much from a simple static asset allocation approach they generally won’t pay much for it either. SAA sounds so much more impressive, even if in practice the two are often much the same thing.

The problem of traditional SAA is that it commits the investor to a particular set asset allocation today based on history (often also heavily influenced by competitor weightings) with little if any forward looking analysis. A new client who comes in a year or two later will likely be given much the same allocation.

But is SAA really a problem, particularly if some only use it as a reference point? Yes, because if you start with the anchor of a SAA then psychologically it is very difficult to significantly deviate from it. It is also very difficult to introduce new asset classes or sub-asset classes or investment areas that only have a small weighting in asset benchmarks.

Further, the approach is also very slow to deal with structural changes in asset classes that can markedly change their risk/return profile.

One of the big debates at the moment is whether retirees have too much in growth assets. Historically, I would agree that this has been a problem, partly driven by remuneration structures that promote higher fee growth investments.

However, because this debate is taking place with SAA as the incumbent asset allocation framework, the current debate is simplistic at best and plain wrong at worst.

For example, is now the appropriate time to be arguing that equities (where arguably many markets are cheaper than any time in the past two decades) should be sold down and replaced with bonds (at their lowest yields in the last two decades).

Indeed, this highlights a clear flaw in ‘Lifecycle’ funds that rely on different SAAs based on age. If we didn’t have the flawed SAA framework as background there would be no choice but to also incorporate long-term valuations and other fundamental drivers into such asset allocation debates.

What about risk profiling? 

Surely incorporating risk profiling requires that clients with particular ‘risk profiles’ need to be directed to specific SAA’s which represent different risk levels. While this has become the accepted approach there is nothing prescriptive in legislation that necessitates a specific SAA as an outcome from formal or informal risk profiling.

In fact, I believe the relationship between risk profile and asset allocation is flawed at both ends. At the risk profiling end, it is flawed because attitude to risk varies over time (in an inverse way) so investors tend to prefer lower risk options after higher risk options have performed poorly and vice versa, a tendency that actually destroys value.

At the asset allocation end, this relationship is flawed because of the inherent problems of SAA described above and particularly because the risk of any specific SAA is not constant anyway.

The key is to determine the level of risk and returns appropriate to meet investors’ objectives over time.

At the investment end we need fewer, but more flexible asset allocation mixes which are more valuation and contrarian driven and targeting the absolute benchmarks – cash plus or CPI plus – that actually matter for retail investors, albeit over the medium-long-term.

Of course, there is no certainty of meeting such objectives, but at least one is starting the journey with a focus on the right destination.

I have long argued that the use of five, six or even seven risk profiles attached to specific SAA’s is pointless overkill.

With this many SAA mixes there is little variation in risk/return between them, but more importantly this approach locks in the flawed SAA approach because it creates a perception that the full range/mix of asset mixes have been covered and the allocation decision can be delegated back to the client – and their risk profile.

At the risk profiling end, I believe questionnaires and assessments should be downplayed and provide just one input into the appropriate structure of the client’s portfolio. Importantly, it is the clients various objectives over time, relative to their investment portfolio’s capacity to meet these requirements that is crucial.

For example, the path of future returns is critical, so retirees should be conservatively invested at first given their inability to regain lost ground in bad markets.

If a wealthy client can meet most of their objectives with a lower risk portfolio does it really make sense to give them a higher risk one just because they show up with a higher risk profile at the time?

Of course, some investors will express such a high level of aversion to risk taking that a riskier portfolio might never be appropriate because they would likely bail out in difficult times, thereby undermining the case for a more aggressive portfolio in the first place.

It is managing client expectations, objectives and cash-flow consideration that most financial planners should be focused on. Implementing a more active asset allocation approach well at the investment end will not be easy for most.

Indeed, some planners and investors will make an absolute mess of more flexible asset allocation and do worse than a static approach. They will chase asset class performance and buy high and sell low if research into investor behaviour is any guide.

However, bad implementation of more active AA approaches should not be seen as reasons to support SAA. 

At one of the PortfolioConstruction Conference’s due diligence sessions on asset allocation using scenarios, the following question was asked: “this is all great stuff but how do we implement it ourselves for clients”?

The answer was handled diplomatically but the proper answer probably should have been 'You can’t', at least not without devoting major resources to the approach, including experienced people. The reality is most financial planners will have to partially or fully outsource more dynamic asset allocation.

While I believe SAA has always been flawed, its weaknesses are being highlighted at a time of dramatic change and challenging investment markets.

For example, traditional safe havens asset classes such as government bonds are being fundamentally reassessed. The role of new asset classes and sub-asset classes such as gold or commodities have needed to be considered.

Equity asset classes are currently dominated by weightings in financial companies which face the challenges of de-leveraging, credit risk and regulatory change. Dramatic price moves can change the prospective long-term returns on a range of asset classes by several percentage points per annum within a matter of days or weeks.

Being anchored to backward looking SAAs will inhibit the ability to deal with these challenges and risk dragging clients underwater.

We need to remove the straightjacket of SAA and move towards a framework that is focused on, and flexible enough to meet, investors’ true objectives. Such approaches are by nature more qualitative and subjective, harder to judge value-add and difficult to implement successfully. 

Many financial planners will find it all too hard and stay with SAA despite its flaws and the feeble rationale for its various inputs and outputs. This would be a mistake.

Their clients might not currently know the difference but I suspect they will increasingly question how their asset allocation is being determined as more progressive parts of the investment industry more openly emphasise SAA’s flaws and eventually abandon it.

Increasingly, there will not be a debate about the merits of SAA versus DAA, but simply a discussion about how best to manage asset allocation in a world that has accepted that the emperor of SAA has no clothes.

In my view the current SAA framework will be largely abandoned by the institutional and retail investment industry within the next 3-5 years. Why wait?