'Asset allocation' is the hardest subject in investment. It is also the most important.

The show so far
In Simple Investing we encouraged you to:-

• Understand the need to take some risk;
• Control that risk by finding a suitable (for you) balance between cash and shares;
• Spread that risk by diversifying.

In More Diversification, we introduced correlation and the principles of diversifying among shares. Now we ask: can we benefit by diversifying among asset classes - expanding our potential list of assets from the two classes of cash and shares?

This process is called 'asset allocation'.

Adding assets
We list nine asset classes in the assets section of this site. These are cash, shares, bonds, index-linked bonds, premium bonds, property, buy-to-let, commodities and collectables. You may be quite happy to stay with the cash/shares mix you have found. Or you may want to try adding a third asset class.

There is nothing magic about the process now. You have to decide how much of your preferred third class to buy and what proportion of cash and shares to replace in your portfolio.

You will be guided by the expected return on your chosen third asset class. To maintain the risk profile you are comfortable with:-

• a high return will indicate high risk and you will want to mostly give up high-risk shares;
• a low return will indicate low risk and you will want to mostly give up low-risk cash.

You should end up with a new portfolio you are comfortable with, and maybe a bit of extra diversification. Against that you have the complication of managing a third asset class.

Adding a bit of science
You may be thinking all this is a bit ad hoc and there ought to be a scientific way of doing all this. and there is. It is called Portfolio Theory.

We have mentioned 'correlation'. The concept can be expressed mathematically, such that between any pair of assets there is a correlation coefficient. If you run all these numbers through a computer and tell the computer what level of risk you are comfortable with, the computer will spit out an investment portfolio (or a family of portfolios) that will give you the highest return you can get for that level of risk. These are called efficient portfolios.

It gets better. It seems to be a characteristic of efficient portfolios that they include a small dash of 'unconsidered' asset classes- that is, classes that do not seem to be particularly attractive investments in isolation.

It is hard to see, intuitively, why this might be so. The best explanation we can give is that the 'unconsidered' asset has low correlations with cash and shares which allows the computer to substitute mostly low-return cash without increasing the risk of the portfolio. But generally, we must admit that if you can't do the maths (and very few can) you'll have to take this on trust.

Which raises the question....

• why don't we entrust our money to a Fund that has the expertise to find the efficient portfolios for us, and/or,
• why don't we use the information provided by fund groups, who may, for example, advise us that their computers show that 5% in some alternative asset class added to a standard portfolio will both increase return and reduce risk - the holy grail?

We spend time on this because these are extremely persuasive sales pitches and you need to understand why you might resist them.

...and the answer is.....
There are a number of of contrary factors:-

• Garbage In, Garbage Out applies. There is a correlation coefficient for every pair of assets in the model. With 10 assets, that's 45 coefficients. That's 45 forecasts of how two assets are going to relate to each other in their future performance. You might think that if someone is clever enough to get that right it would be easier just to forecast future returns and pick the highest.
• The maths assumes that correlation coefficients are fixed over time. There is no reason why that should be so. In fact there is now some evidence that correlations increase in bear markets - in other words, diversification benefits break down just when you need them.
• There's a management trap called 'Managing to a Model' and you are in danger of falling into it. Portfolio Theory constructs a model of the investment problem and optimises it. The results provide insight; but they don't provide answers. It's only a model - a simplification. It uses just one definition of risk, for example.
• As always, you have to consider the cost of taking advantage of a perceived benefit. The sorts of improvement we are looking for here are maybe an extra return of 1/2% for the same degree of risk. That is well worth going for over a long period (compounding); but only if the costs of doing so are less than 1/2% per annum. Which they won't be.

Do it with shares
You can get quite close to some asset classes by buying shares. If you want some exposure to property, buy property shares. If you want some exposure to commodities, buy mining shares. If you want some exposure to corporate bonds, buy gilts spiced with a few equities.

Other ways of slicing the cake
A reminder to finish with. It is easy to get hung up with asset classes. You can slice up the whole universe of investible assets in any way you like. Every way gives you a different way of looking at things, a different set of decisions and, if you are not careful, another way of persuading you to lose the wood for the trees.

Stay with first principles - you need to invest in a range of different stuff, but the benefits of diversification fall quite quickly, and the complications increase, as you add more stuff.

This is the most important investment decision you have to make. If you get that right, you are 95% of the way towards a proper savings plan.

What assets do I chose from?
There is no commandment that says assets must be grouped in a particular way. But it has become conventional to focus on four asset classes: cash, property, bonds and shares. Most savers will restrict their investments to these four.

Commodities and collectables are two other groups worth mentioning. Though we don't recommend them.

Can't we divide groups of assets with finer cuts?
Well, yes. We don't really want to do it, because we think you should keep it simple. But we will mention some common subdivisions:

• In property: your own house(s), retail ('buy-to-let') and commercial.
• In bonds: UK government ('gilts') and 'other' (foreign government, investment grade corporates through to junk bonds). Perhaps the most important: index-linked bonds (a subset of gilts)
• In shares - too many to mention.
• In commodities: gold and 'the rest'.

What do we do now?
Get some specific advice. And the best of luck. Ron Sandler (he of the influential Sandler Review), in his submission to the Treasury Select Committee (News Bites), opines: "there is almost universally no attention given to asset allocation by advisers in this country. Their focus is on product purchase."