Let every man divide his money into three parts, and invest a third in land, a third in business, and a third let him keep in reserve. - The Talmud, Circa 1200BC-200AD

A useful simplification of Modern Portfolio Theory

There are a number of criticisms that can be levelled at MPT, but it would not be a good idea for you to give up on it altogether. When you strip away the over-complicated mathematics and excessive reliance on past correlations that won't hold in the future you do find that there are some very useful concepts left in there.

A book I read recently that crystallised my thoughts on the subject was The Intelligent Asset Allocator by William Bernstein. I suggest that you should probably read this book yourself, as it is not a particularly challenging book in terms of hard mathematics or anything, but does offer a fairly state of the art model for portfolio construction.

So what are the main problems with Modern Portfolio Theory? There are lots of them, but the big ones are:

• Volatility is not necessarily a particularly good measure of risk, particularly for individual assets such as shares. Even if it was a good measure of risk, volatility varies from one period to another and thus can't be anything more than a measure of risk in a historical period.

• Past correlations are largely a product of coincidence, and probably won't hold in the future. In particular, at the very time that you want your carefully designed portfolio to hold together, at the time of a great crisis, most things will suddenly become highly correlated as various sectors all move down in unison. You should not put too much faith in an "efficient" portfolio performing at all well if world markets go mad for a little while.

• Because volatility, risk and correlations do not behave in the same way from one period to another, the concept of the "efficient frontier", while apparently true, is not useful in practice. If you get a computer to chart risk and return for a variety of portfolios, you do get an upward limit on return for a given risk, that does follow the "efficient frontier" left to right convex curve. Although empirically we have found that efficient frontiers do exist, there seems to be no way of predicting in advance what asset allocations would give you a portfolio that lies on this curve. Therefore people that set a computer to carefully plot out a thousand portfolios to find which asset allocations have worked, will get one that only worked up until yesterday, and this portfolio will behave differently in the future.

Ok, so all that maths and number crunching that people do to find ideal asset allocations seems to be largely a waste of time, but what are the basic features of MPT that are useful, and how do they benefit us?

Well simplistically, the idea behind MPT is that if you have two assets with a similar return, but a very low correlation, frequently when one asset zigs, the other zags, and as such you can create a portfolio with more stable returns by putting 50% of your money into each of them.

So far so good, as long as you aren't trying to get too clever and model the perfect portfolio you won't go wrong by investing in two different assets, in particular if they have similar over-all returns.

What are some examples of this sort of portfolio? Well, contrary to what fans of either asset class may tell you, the long term returns of blue chip commercial property trusts and diversified shares portfolios have been quite similar. They are correlated to an extent, in that interest rates tend to have the same effect on each, but the correlation of property trusts to, say, the All Industrials Index is not particularly high. Given that shares are going to do well in low inflationary times versus property which will do well in times of high inflation, an argument can be made (irrespective of whatever a computer has found to be "optimal") that a portfolio made of 50% property securities and 50% Australian shares would in all probability have a similar return to either asset class but with lower overall risk.

What you are looking for are assets that will offer a similarly high long term return but with a low correlation. There is little point putting any money into a very poor returning asset just for diversification reasons, unless you have a special reason to be risk adverse a major sacrifice to returns is not sensible just because it brings down volatility a bit, so what you should be looking for should be a portfolio of high returning assets provided that you can buy a bunch of very different ones.

What about international shares versus Australian shares? Well, the long term performance is probably about the same, give or take, but historically the correlation between the Australian market and most world indexes such as the MSCI have not been too high. I might argue that whatever portion of your portfolio that you have dedicated to shares should be split in half and allocated between international and domestic shares.

The international shares portfolio will do quite well if Australian inflation gets out of hand, causing our currency to depreciate against international currencies. Therefore you might well want to put some implicit inflation hedging into your portfolio by buying a non-hedged international fund. Of course if you want to put half of your international money into a hedged portfolio and half into a non-hedged portfolio, that will create more diversification, without necessarily having any great long term impact on returns.

If your shares portfolio is split into several sub-classes, maybe you could allocate more than half of your money into shares, reducing your property exposure? Although not a view shared by everyone, I personally like the idea of a portfolio made up of:

30% International Shares

30% Australian Shares

40% Property Securities

(Where the property securities is a property managed fund, a property index such as the Vanguard Property Securities Index fund, or just a few shares in real estate companies or listed property trusts).

So 60% of your portfolio is shares, which would do well when inflation is low but the economy is doing ok, and 40% of your portfolio is dedicated to a somewhat more "defensive" asset class that holds its value well (in the long term) when inflation is high. Most people I know would protest that this is far too much property to be holding, favouring anything up to 90% shares, but I commonly overweight portfolios in property, especially if gearing is being used, because it reduces volatility - something that shouldn't bother a long term investor but one needs to bear in mind anyway if you are gearing. This allocation would be my personal choice if I was making a geared portfolio up out of large capitalisation index tracker funds and makes sense to my way of thinking. A more complex portfolio would include value indexes and small capitalisation stocks to a greater extent, read the article on "entire market" indexing.

William Bernstein makes up portfolios out of dozens of asset classes, with use of various "value" and "growth" funds, smaller companies indexes, large companies indexes, bonds, cash, property trusts, and various other things. He describes himself as an "asset class junkie". If you want to get really serious about asset allocation, you can find all sorts of different asset classes if you want to look around.

In the section on managed funds, I disparage "index huggers". An "index hugger" is an active fund that maintains a portfolio largely similar to an index, but with small changes where they see fit. (Not to be confused with an "index tracker", which is a non-active fund that doesn't pretend to be active and hence charges a low fee, lower than an index hugger, given that they never said they would try to outperform the index.)

Here is one of the places where you will see why I don't think index huggers are worth buying. The reason why you want to buy an active fund is to give you performance that is very different to that of an index. Even if all they do is perform as well as the index, you want them to perform as well as the index with a low correlation. Why buy an index hugger when you can just buy an index tracker, and get the same correlation and probably the same returns?

What you want, if you make the decision to buy an active fund, instead of an index fund, is a fund that ridicules the notion of trying to minimise their "tracking error", one that has a chart that looks absolutely nothing at all like the index it is benchmarked against, which will provide additional diversification. If you buy five index hugger funds you'll get exactly the same performance as an index tracker, but with more complexity, less tax efficiency, and more fees.

How would you fit such a fund into your portfolio? Well lets just say that there was an active international fund that despised conventionality, considered short selling whenever they thought some share was overpriced, but the rest of the time took long positions in relatively obscure companies that aren't even in the index? (I am not saying you should or should not buy a fund from this manager, but an example of such an unconventional approach is followed by Platinum, and the same comment could be made of several other managers, as well as many other equity hedge funds and long/short managers.)

In this case, you might make a portfolio like this:

20% unconventional international fund or international long/short value fund

20% MSCI index tracker

30% Australian index tracker

30% Property Securities index tracker

If you want to split, say, the Australian Shares asset class in half, and provided you can find a value fund and a growth fund that you have sufficient confidence in to favour over a style-neutral index tracker, you might want to buy a dedicated Australian value fund and a dedicated Australian growth fund, in equal proportions, each representing maybe 15% of your portfolio.

After that you can put a certain amount into small cap funds, theme funds like a health care fund, hedge funds, this is what diversification is really about.. a diverse range of investments.

You could split up the property sector with a direct purchase of a residential property in addition to a property trust, if you wanted to, but perhaps a better option would be to look for an international property trust, or a property trust that is "different" in some way.

Bearing in mind the Fama/French Three-Factor Model, we might decide that small company stocks and value stocks are the way to go, so we increase our exposure to these. You may end up with a portfolio something like:

10% Australian market index

10% Australian value stocks

10% Australian small company stocks

5% Australian small company value stocks

10% MSCI market index

10% international value stocks

10% international small company stocks

5% international small company value stocks

5% emerging markets

25% real estate trusts

There is a lot of room to change these asset allocations around, all I've done is divide a portfolio into roughly equal parts to include major asset classes, so don't get the idea that there is something sacred about the weightings suggested, you can go much further in making a portfolio. You might reduce your exposure to one or two of these asset classes and put a bit more into the previously mentioned unconventional long/short manager. This is what I do when constructing portfolios, you can split your portfolio into index vs active managers, or do your own stock picking for Australian shares. There are a number of good active funds out there that I like, I slot these into the portfolio to create any desired level of risk and return.

If you were a more risk adverse short term investor you would drop the emerging markets and small companies and replace them with cash, short term bonds, good quality mortgage trusts etc, in fact you might make these conservative asset classes represent much more of the portfolio. If you read my article on income planning you might get some idea of how much conservative assets could be used for a portfolio designed for producing an income.

Well, all this diversifying would certainly annoy Warren Buffett, who as I have said already is a dedicated "focus investor", preferring intense concentration on exceptional assets, without regard to any kind of strategic portfolio planning. While no one using this broadly diversified MPT approach is going to get returns that come even close to Warren Buffett's 30%pa career batting average, the advantage of the MPT approach is that you don't have to be history's greatest investment genius to pull it off with great success.

The next section of this FAQ, on contrarianism, explains partially how MPT can actually create returns that are superior to the sum of the individual assets' returns, even while reducing overall risk.

The contrarian effect seems to show that you will do well by buying investments that have performed really poorly over the previous year and selling last year's best performing assets (read the contrarianism section for details).

The contrarian effect can provide a boost to your portfolio returns because you will be constantly allocating more money to last year's worst performing asset class, partially selling out of sectors that have boomed. You ought to be able to wring perhaps another percentage point a year of performance out of your portfolio like this, but almost as importantly the contrarian sector tends to have lower downside risk.

Portfolio rebalancing, the process of resetting your portfolio at predetermined intervals back to a default asset allocation, is a form of contrarian asset allocation and is performed automatically by most "balanced" managed funds. In the next article I will expand a bit more on strategic and tactical asset allocation strategies.

If you are not what is referred to as a "High Growth" investor, that is you desire a more stable and income focused portfolio than one with 100% allocation to growth assets, you might use the above asset allocations for the 85%, 60%, 30% or whatever of your portfolio that you have dedicated to growth assets, allocating the other 15%, 40% or 70% to bonds, mortgages or cash.

One fortunate factor that does make matters a little easier for you is that asset allocation does not have to be too precise in order to work. Bernstein claims in his book that allocations can be off by 20% either way, giving fairly similar long term returns. The most important factor is not what asset allocation you choose, but that you stick to it and rebalance your portfolio as required. The mere act of rebalancing is what gives you the higher returns with lower risk, not necessarily the asset allocation strategy itself.