Portfolio management in the real world

Life is messy, and investing is no exception. Dominic McCormick writes why he believes investors are not well served by current asset allocation methods.

The global financial crisis has rightly sparked debate around issues relating to portfolio construction and asset allocation.

In particular, there are questions about whether the conventional, largely fixed, strategic asset allocation (SAA) approach makes sense for many investors.

I have always struggled with definitions of SAA. They are usually couched in vague terms such as your ‘normal’ or ‘neutral’ allocation, or the asset allocation that investors expect will meet their long-term risk/return objectives.

Usually they are developed with a heavy reliance on long-term historical returns and risks.

The problem is that in an inefficiently priced world, where bubbles and busts occur, and markets can and do get significantly over and undervalued, none of these definitions make sense.

This is particularly the case in a world where even ‘long-term’ investors’ true practical time horizons are often as short as a few years.

In such a world, there is no pre-determined, fixed SAA that one can be highly confident will meet an investor’s return and risk objectives over their investment time horizon.

For example, it has been frequently demonstrated that starting valuations are a hugely important driver of future returns of asset classes, even over extended time periods, yet current valuations are ignored or heavily downplayed in most SAA determinations.

The reality is that the risk premiums earned from asset classes are not constant and reliable over time. Instead they vary, sometimes dramatically, even over the five, 10 and 20-year periods that are typically considered as long term.

Just look at the US or Japan where any standard asset allocation with a fixed high weighting to local equities has failed to deliver positive returns, let alone meet any reasonable absolute return objectives, over the past decade.

The only world where the traditional SAA approach makes sense is one where the Efficient Market Hypothesis (EMH) holds and markets are fairly priced all the time.

While only a fading cult of diehard adherents still believe in market efficiency, many more still cling to its offshoots such as Modern Portfolio Theory and SAA.

What about tactical asset allocation (TAA)?

Isn’t this the way that investors can deal with some of the limitations of SAA, by being prepared to tilt asset allocation away from the SAA within defined ranges?

In theory perhaps, but in practice TAA has usually been disappointing for a variety of reasons. It is typically a very narrow strategy only dealing with a small number of possible bets (eg, shares versus bonds and property trusts), and usually within narrow ranges.

For most of the 1990s and 2000s, when these assets were all in bull markets most of the time, the scope to add value by shifting between them was clearly limited.

TAA also tends to focus on those harder-to-make, short-term moves (since SAA is supposedly looking after the long term) and there is often pressure to be seen to be active in making TAA moves even at times when high conviction opportunities are not present.

TAA around a flawed concept — SAA — is simply flawed itself.

How sensible is it to make small TAA tilts (usually on a valuation basis) away from the more dominant SAA, which itself has been determined (or is fixed) with little, if any, reference to valuation?

TAA as practised often ends up as a half-hearted attempt to compensate for major flaws in the SAA process by minor fiddling. It is no surprise that it usually fails.

The investment industry is slowly recognising the problems with SAA and TAA as practised.

Several major consulting groups have introduced what they call “dynamic strategic asset allocation” or “strategic tilting” with the rationale being to be able to occasionally diverge significantly from an investor’s SAA when markets get dramatically over or undervalued.

This makes sense, but I don’t think it goes far enough and I am not sure the labelling is correct. If it means you can alter the SAA at any time (even if infrequently) then clearly it’s not really strategic.

Some have simply moved to reviewing their SAA more frequently, say every couple of years, but this approach is arguably neither strategic nor flexible.

It seems that after espousing the virtues of SAA for the past couple of decades, some cannot let go of the SAA framework despite its increasingly obvious flaws.

Ultimately, all that really matters to investors is their current, or target, asset allocation, not whether it is strategic or tactical.

This asset allocation is the one that will earn the investor’s returns and expose them to various risks.

Handing the bulk of that asset allocation decision to history and failed investment dogma, as the SAA approach does, is lazy at best and downright dangerous at worst.

A better approach, in our view, is to abandon fixed SAA benchmarks altogether in favour of wide asset ranges and an increased flexibility within those asset classes.

This allows for asset allocation moves (and very large ones if required) without the pressure to always consider small tilts under a TAA approach.

One might still decide to retain a benchmark asset allocation mix for longer-term performance comparisons only, but this mix should not in any way influence asset allocation positioning.

This is not to suggest that abandoning SAA and implementing a more flexible and proactive approach to asset allocation is easy. It clearly is not.

To add value requires a combination of discipline, the ability to take contrarian views and patience. But in my view, you have no choice.

The few remaining believers in the EMH may beg to differ, but there is no passive alternative in asset allocation as there is with stock selection.

Even adopting a particular SAA is an active decision to allocate to certain asset classes, albeit a heavily constrained, inflexible and narrow one.

You are kidding yourself if you think you are not making a judgment on long-term valuations when you chose a particular SAA.

These concerns over SAA also have implications for client risk profiling, which typically has a number of different standard SAAs as the solutions for various risk categories.

Clearly, if you accept the view that a certain asset allocation will provide widely varying exposure to investment risk over time as valuations and asset structures change, then tying a certain risk profile to a fixed SAA is clearly a mistake.

Only an asset allocation process that is prepared to reduce exposure to an asset class as its risk increases (and vice versa) has any chance of keeping risk reasonably constant over the investment time frame of the risk/return objectives.

This industry has done clients and planners an enormous disservice over the years by pretending that investment is much more scientific than it really is.

It is this ‘physics envy’ that enabled flawed paradigms like EMH, Modern Portfolio Theory and the SAA/TAA construct to become so entrenched within the investment industry.

The behaviouralists are bringing some balance back to this discussion, although they haven’t (and in my view won’t) come up with an all-encompassing model of the way the investment world works.

We need to accept that successfully navigating investment markets involves dealing with uncertain fundamentals, making subjective judgments, managing behavioural biases and adopting commonsense.

Quantitative inputs and academic theories are tools to assist in making decisions, not fail-safe constructs to tell us how to think.

This approach may lack the precision we all crave but it does encourage us to make decisions based on the way the world really works rather than how we would like it to work.

Or how it has worked during what is often very unreliable history — and history that, in any case, usually spans a far longer time frame than that which most investors are expecting to achieve their risk/ return objectives.

Dominic McCormick is chief investment officer at Select Asset Management.