The ‘science’ of asset allocation

Traditional asset allocation could be defined as offering strategic allocations covering the five to six major asset classes and typically offering three to six client asset mixes with different equity exposures for each risk profile.

Most strategic asset mixes seem to be developed to look broadly similar to industry competitors and are often heavily derived from 10 to 30 years of asset class history. There is usually no planned adjustment of these strategic allocations over time, although some minimal tactical asset allocation may be undertaken.

In previous articles, I have suggested an alternative approach and highlighted some limitations of the existing approach. I have my share of critics — including Graham Rich who recently said I misquoted the Brinsonet alstudies from 1986 and 1991 by referring to those who promote Brinson’s view that asset allocation determines over 90 per cent of portfolio return.

Although I haven’t directly quoted Brinsonet al, I do believe the studies have been widely misused to overemphasise the importance of asset allocation. The Brinson papers deduced that asset allocation explained 91.5 per cent of the “variability” of returns among funds. This is not the same as explaining 91.5 per cent of the level of returns (and it is this that most investors are interested in). Furthermore, the Brinson study covered large (and generally very inflexible) pension funds.

Research conducted by Ibbotson and Kaplan on the other hand, raised the question about whether asset allocation explained 40, 90 or 100 per cent of performance.

A key finding was that the difference in the level of return between funds explained only 40 per cent by asset allocation, clearly refuting the 90 per cent plus Brinson dogma and suggesting asset allocation is not nearly as important as many would have us believe. It is worth noting that the Ibbotson and Kaplan paper went beyond pension funds and looked at balanced mutual funds as well.

How important is asset allocation? My logic is very simple. Using index funds, asset allocation will determine 100 per cent of return. Use high tracking error managers and it might reduce to 50 to 70 per cent. Use hedge funds and it might be as low as 20 to 30 per cent. If this ‘hedge fund mantra’ that is so derided is about generating absolute returns with less dependence on traditional assets classes having to go up as much as history, what is wrong with this?

In moving away from traditional portfolio construction approaches, the importance of asset allocation can be lessened, although other factors and risks (sector/style strategy and stock selection) become more important.

Downplaying asset allocation and adopting a more bottom-up approach is therefore valid. You still end up with see-through asset allocation, but it is wrong to suggest its importance is not less than a traditional portfolio with the same asset allocation.

Graham Rich recently said he is not against building better approaches but only when we “build them scientifically and test them thoroughly. In the meantime let’s apply what’s known to work — a quality multi asset class investment strategy”.

Does traditional asset allocation really work in practice, as its advocates claim, particularly the way most are implementing it today? What has scientific asset allocation done for a client’s portfolio lately or over the elusive long-term? What will it do for clients looking forward?

I recently looked at the performance to January 31, 2003, of several multi-manager retail diversified balanced and growth funds that tend to adopt a traditional asset allocation approach. Three-year returns were slightly negative, while five years were all one to two per cent less than cash. For seven and nine years, the returns ranged from 4.8 to 5.8 per cent per annum, similar to the returns from cash management trusts. Similar figures are derived when you simply take the index returns, assume a standard balanced/growth allocation and subtract a reasonable estimate of retail investor costs.

Does this indicate something that is working? Over all periods, investors would have done better or no worse in cash or a term deposit, although some tax advantages would improve the situation marginally for the traditional asset allocation.

But, I hear some say, these figures are distorted because sharemarkets are close to a bottom. Perhaps, perhaps not.

Markets may not be expensive anymore, but the world as a whole (the MSCI index) is still not cheap compared to history, and the one-off boost to bond and property trust returns from falling interest rates has already happened. We also need to remember that the five, seven and nine-year figures contain two, four and five years respectively of the biggest bull market in history for many global sharemarkets. Is this working? Others say go back longer periods than nine years and you can show it works. Maybe, but how long is long-term?

However, I still believe that some long-term strategic approaches to asset allocation (with regular rebalancing) can work for some investors and/or for part of a portfolio. However, such an approach should skew the portfolio towards assets or sub assets that have sound fundamental reasons for inclusion in the portfolio. This contrasts to the traditional asset allocation approach that simply buys the accepted index for each asset class (or a collection of managers that generate index returns) generally in proportion to what is the accepted or popular industry asset exposure today.

In fact, a combination of a long-term strategic rebalancing approach with the more active approach I am proposing can make a lot of sense. On the other hand, it may well be that the conventional approach gives you the worst of both of these worlds without any of the benefits.

Of course, the believers in traditional asset allocation feel they don’t even have to think about either of these alternatives. After all, they are applying what they know works! No need to consider alternative assets, no need to make longer term adjustments away from popular, expensive assets and towards unpopular, cheap ones (I’m not talking about three months tactical fiddling here) and little need to consider whether sector or style exposures within asset classes should not be a long way from the neutral position that most seem to target.

In some respects, traditional asset allocation has become a game of hope. Hope that equities will outperform other assets, hope that mainstream assets will return the same 10 to 25-year historical returns the industry has been showing clients for the last decade. With the one-off boost from falling interest rates gone and share market valuations still not cheap in a long-term sense, such hope appears misplaced.

Of course, many traditional asset allocators are under the misconception that investment is a science. Unfortunately, investing is not, and will never be, a science. It is perhaps put best by Dr Jim Walker in the foreword of Mark Faber’s most recent investment book,Tomorrow’s Gold,in discussing Faber’s investment approach.

“This framework eschews formalised mathematical models and relies instead on deductive logic and an understanding of human nature. It is necessarily subjective, but that — whether the quasi-scientists in the economic mainstream like it or not — is reality. Ultimately, greed and fear are at the heart of most, if not all investment decisions.”

Orwell’s quote “Each generation imagines itself to be more intelligent than the one that went before it and wiser than the one that comes after it” does not mean that the conventional wisdom should not be challenged. Progress is made by independent thinking and the willingness to question traditional practices and beliefs.

Of course, history is still a great teacher and the base from which knowledge can be developed. Knowing financial market and investment history (both theory and practice) is vital to truly understanding investments.

As for the future, no one knows accurately where investment theory or practice will take us, but it makes sense to spend a lot of time thinking about it. You are then in a position to realise the limitations of current practice and how much more we need to know in this truly dynamic area.