View more on these topics

AA breakdown

To the 6,000 ARs who were working for Prudential back in 1991-93, I owe you an apology. Actually, I should only apologise to the several hundred people who came on the asset allocation training courses I ran for Pru but I know the training material was copied and widely used by others.

Anyway, I apologise. I told you asset allocation was the only foolproof method of lowering risk in portfolios to a predictable level. I told you lack of correlation between asset classes like bonds, shares and property reflected their fundamentally different economic characteristics.

I made you create “balanced” portfolios using standard asset allocation methods (standard now, but new in those days), and you went on – in many cases very successfully – to sell the concept to clients. I was wrong.

This year, those balanced portfolios have tanked. Even a cautious portfolio containing 40 per cent equities, 40 per cent bonds and 20 per cent commercial property is down 25 per cent, and even worse, this cautious portfolio is down almost as much as many “riskier” portfolios.

So asset allocation (AA) methods have failed – comprehensively. It is easy to dodge this, so remember, asset allocation is not fundamentally about increasing return. To increase return, follow Warren Buffett: “Diversification may protect wealth, but concentration builds wealth.” Clients know this. AA-based invest-ment advice is primarily focused not on gains but on reducing risk. As an investment adviser, that is why you have failed – you did not reduce clients’ risk as much as you said you would. The key question now is – what do you do next?

The institutional response to the failure of conventional investment planning methods runs like this:

The 2008 crash is a once-in-a-lifetime event. It was unpredictable (unpredicted by all the central banks, Treasury departments, etc). Downside correlation in a crisis is inevitable but will not last. Asset allocation methods will work again when things return to normal.

This amounts to – there is not a problem. Carry on.

As an investment adviser, you need to understand that this response is not only inadequate. but is also wrong and will cost you business.

Asset allocation methods are based on work of Harry Markowitz in the 1950s, which generated a theory that says risk can be controlled. The reason it can be controlled is investment returns fall in a predictable pattern, the bell curve or Gaussian distribution. Standard deviation, the proxy for risk in this theory is a mathematical technique only valid when applied to Gaussian distribution. If investment returns do not form a bell curve, SD does not work. And they don’t.

Some modellers weasel round this with the argument that returns do not vary that much from bell-curve distribution. In fact, not only are the chances of extreme events significantly greater than SD assumes, the scale of losses in extreme events is also much greater. The problem with SD is that if you say “68 per cent of the time, returns will fall within one standard deviation”, the intelligent client should ask: “And what of the other 32 per cent?”, a question to which conventional theory does not have an adequate answer because the range of returns given by the model patently does not accord with reality.

Since the crash of 1987, scepticism has grown about the validity of the theory and recently attained popularity in Nicholas Taleb’s The Black Swan. Taleb’s work is discounted by academics because he is a trader and for the real deal you have to turn to the man who laid the foundations for the theory in the first place, Benoit Mandelbrot, the mathematician who invented fractals and chaos theory and whose student Eugene Fama invented the “efficient market hypothesis”. Mandelbrot’s The Misbehaviour of Markets’ does a full-on demolition job on conventional portfolio theory.

It is now accepted in finance theory that, as a matter of fact, investment returns do not form a Gaussian distribution. Banks use GARCH, VAR and other tricks to measure risk and do not rely on SD. But these tools are only sticking plasters, they do not change the wrongness of the theory, they just work round it. In turn, they are being found wanting, particularly VAR, which is now discredited as a risk management tool.

Portfolio modelling tools use mathematical techniques based on the theory, including SD. Some tweak the maths to allow for kurtosis but they still end up assuming the chances of extreme events and the size of losses arising from them are lower than historical fact. For that reason alone, these models are dangerous.

Now turn to the other bias we have introduced to investment planning – expectations. For over 20 years, academics have ground out the message that the equity risk premium will be lower in future than in the past, that we will earn no more from shares than an annual 2-3 per cent above government bonds. That conclusion, too, is based on an axiom of conventional theory that has been proved false – it assumes investors are indifferent between profit and loss, whereas we know that investors are far more risk-averse than gain-seeking and that this consistently biases their behaviour.

If we assume the equity risk premium will be low, we have to put more into equities to get a desired rate of return. So we end up putting more capital in risky assets – which we say is OK, because our models say shares are less risky than they in fact are.

This is a highly condensed account of why the conventional AA model is broken. What should we do about it?

First, abandon any trust in portfolio modeling tools. If you did not understand their maths, you should not have been using them in any case.

Second, assume a level of risk in real assets for the future similar to what we are experiencing now. Don’t buy the one-off blip argument – it does not fit the facts.

Third, assume future returns from risk assets will be in line with long-term historical returns (about 6 per cent above government bonds) or even higher because risk aversion will be higher than normal for the next decade.

This will drive you to a higher allocation in all portfolios to cash and short-term bonds. American practitioners suggest that a “new AA” model could use a set of valuation metrics to govern the switch of some of the cash into and out of risk assets.

Fourth, develop your own message about this shift in AA strategy and deliver it consistently to clients along with any necessary portfolio changes.

Address the issue like this and you may preserve the faith and capital of your clients. Shrug your shoulders, pass the buck and hope for the best and you can expect, and deserve, to see your clients walk. They will not tell you they are going to walk, any more than tens of thousands of stockbrokers’ private clients did in 1974. But unless you give them good reasons not to, walk they will.

Chris Gilchrist has been a financial journalist since 1971. He has also worked as an investment adviser and consultant to financial services companies and was a member of the founding management team at The Burns Anderson Indepen-dent Network. He currently edits an investment newsletter, The IRS Report, and is a director of an IFA business, Churchill Investments.



Salve for the sole trader

Sometimes even a tough Lancashire lad like myself has to admit to being very worried.

Woolie thinking

So, farewell Woolworths. The really sad thing about your passing is that so few of us were surprised. As a working example of the way that retailing has changed, you exemplified how not to survive. Woolworths was, of course, closely followed by MFI in the closure stakes.


News and expert analysis straight to your inbox

Sign up


    Leave a comment