Expert Feed
Poor arguments do not lead to better investment decisions
BY JOHN DYALL | MONDAY, 9 JUL 2018   11:30AM

For all the good and challenging thought in the investment industry it has its fair share of opinion pieces and marketing collateral that look important on face value, but when you dig a little deeper don't quite make sense.

I've come across a few examples recently.

The first is the inappropriate use of the efficient frontier. The efficient frontier is a foundational concept in portfolio construction. Whenever anyone talks about holding an asset for diversification purposes they are referencing the efficient frontier. Fundamentally it means holding a portfolio of different assets that all have positive expected returns should lead to a better return per unit of risk (using volatility as a proxy for risk) than holding single assets or single asset classes.

Importantly, it is based on expected return, expected volatility and the expected returns relationship between assets. The examples I have seen all show historical returns, historical volatilities and historical correlations between assets. And they show this over a single limited time period that invariably (one might say inevitably) leads to the conclusion that the adviser should incorporate the product being discussed immediately into all client portfolios.

But it also throws away one of the important assumptions behind the efficient frontier: that returns are normally distributed. A normal distribution has a mean (the expected return) and a standard deviation (volatility that doesn't change much), it doesn't lean too far to the left or the right (called the skewness of the distribution) and it has tails that aren't too fat (meaning there are a lot more extreme returns than you would expect) or too skinny (meaning that returns are pretty much all between a lower limit and a higher limit).

The efficient frontier examples use historical returns (which invariably look good for the promoter) and make no mention of whether the returns are normally distributed or not. It's just a bad argument falsely presented.

In another example the investment manager has put out a white paper to pursue an argument to promote a product that consists of a portfolio of equally weighted Australian shares.

Now I am all for equal-weighted portfolios, particularly in the absence of any other information. In Australia they make particular sense because of the size of the financial sector, which is dominated by the bank mega-stocks. Having less of them in your portfolio reduces the common risks they all face.

The argument put forward by this manager was both complex and (to me) confusing. It appears to be based on two pieces of information. The first is that the lowest return you can get for a single share is -100%. That is, the company goes broke and all the equity holders get nothing.

The second is that the returns from the shares of smaller companies are skewed more to the right than those of larger companies. In other words, smaller companies have the potential for much greater returns than is possible with larger companies.

This makes sense because all companies participate in the real economy. The larger they are the more they come up against the speed limit of growth imposed by that economy. Smaller companies can grow faster because they have less of an impact on the economy than larger companies.

A large company that had the same probability of extraordinary returns (say 1000%) as a smaller company would soon take over the economy. But does it then follow that an equally weighted portfolio is the most efficient? I couldn't come to the same conclusion as the manager from reading its research.

Despite a detailed and comprehensive effort, it's my belief they lost the opportunity to provide a simple cogent argument in favour of their investment approach.

This opinion article was first published in the July 9 print edition of Financial Standard. You can also view it on our free iPad app.

Link to something Infj65w5