Past is imperfect for those looking ahead

PREDICTING the past is easy - everyone who's wise after the event does it. The problem with predicting the future, on the other hand, is that most of it hasn't happened yet. But it's the past to which financial planners and fund managers turn when trying to predict how investments are likely to perform in the future. And that can be a problem for investors.

Tim Farrelly, principal of Farrelly's, a specialist asset allocation consulting firm, says using past performance as a guide to likely future performance can be dangerous. But when we make educated guesses about the future, the past is all we have to go on.

The past per se is not necessarily a misleading guide to the future, Farrelly says, but real trouble stems from assuming an asset class or investment market will perform in a particular way in the future because that's how it's performed in the past.

Unfortunately, it's precisely this assumption that investors often make. And so do financial planners when it comes to the task of determining an investor's asset allocation, or how to spread an investor's money between different asset classes.

However, "if you look at past performance and understand how you achieved that, you can ask yourself how likely are those things to happen again", Farrelly says. It then becomes obvious why investment markets are unlikely to repeat their most recent 10-year performance over the next 10 years.

Farrelly says it's easier to predict long-term than short-term returns. Short-term returns tend to be far more volatile and can be affected by transient issues.

Nevertheless, using past performance to forecast the future can still be misleading if you don't understand the different components of investment returns.

As the accompanying chart shows, if you assume the next 10 years are going to be the same as the 10 years just past, then you will generally be in for a surprise - on the upside as well as the downside.

The chart has two lines - a red line and a blue line. The red line shows the forecast you'd have made if you'd used assumed the previous 10 years would be repeated. The blue line shows the actual 10-year performance that occurred for that same period. The chart shows that the actual return - the blue line - is almost never in agreement with the forecast based on the past, the red line. In fact, it's almost a perfect mirror image.

The chart shows that in the early 1970s, for example, the red line was up around the 15 per cent a year mark, meaning that if you'd used the previous 10 years as a guide you'd have expected 15 per cent returns for the next 10 years. The blue line says that you actually got returns of less than 5 per cent a year.

Conversely, if you'd assumed the 6 to 7 per cent returns of the late 1980s were going to be an accurate guide to the future you'd have been (pleasantly) surprised to have actually received returns of around 15 to 17 per cent a year instead.

Farrelly says a headline performance figure can be profoundly misleading, but if you look behind the numbers you start to see where returns come from, and begin to understand why past performance generally can't be repeated indefinitely.

He says there are three components that determine an asset's return over time. The first is the income you get from it today, the second is the rate at which the income is expected to grow, and the third is any change in valuation multiples, such as the price/earnings (P/E) ratio for shares.

The P/E ratio is the share price divided by the earnings per share. So the higher the P/E, the higher the value of the share compared to the dividend, and the lower the P/E the lower the share price compared to the dividend. High P/Es tend to suggest that shares are becoming overvalued.

Let's say the dividend yield from an Australian share is 4.5 per cent, and earnings for that share (and hence dividends) are expected to grow at 4 per cent a year. If the P/E multiple of that share does not change then the return will be 8.5 per cent a year.

But if the P/E ratio moves, the return will be affected - because the P/E ratio changes as the share price increases or decreases compared to the dividend.

Say the P/E ratio moves from 16 times to 18 times over a 10-year period. That's equivalent to an additional return of 1.2 percentage points a year, so the return from the share will increase to 9.7 per cent a year.

Conversely, if the P/E ratio were to decline from 16 times to 14 times, its effect would be to reduce the share's total return by 1.3 percentage points a year, dragging it down to 7.2 per cent a year.

Understanding why investment markets can't perform at a given level indefinitely has implications for one of the most crucial steps in investment planning, and that is asset allocation.

Farrelly says asset allocation is the link between an investor's goals and aspirations and investment strategies. And he says goals are best expressed in cash flow terms (such as: "I need $50,000 a year when I retire" or "I'd like $50,000 a year but I don't want less than $30,000 a year").

Asset allocation is the name for the process of spreading your investments among different asset classes. It's generally accepted as a tool to control the amount of investment risk you're exposed to. Farrelly says "risk" needs to be defined as the probability of an investor not meeting their cash flow requirements.

The traditional approach to asset allocation is based on an acceptance of what's called the "efficient markets hypothesis", which holds that investment markets accurately price all assets, all the time. But markets aren't always efficient, and the danger is that investors get sucked into buying risky assets just when their returns are likely to be the lowest.

Using forward-looking forecasts, rather than assuming the past will be repeated, helps to construct a portfolio with particular cash flow objectives and risk objectives.

Farrelly also says portfolios should be constructed with reference to an investor's personal preferences (they may want to avoid investing in companies that are involved in gambling or tobacco, or manufacturing weapons, for example) and as tax efficiently as possible.

He says forecasts should be reviewed regularly, and portfolio construction should be based on these forecasts. A portfolio based on today's forecasts might look different from a portfolio based on forecasts made in a year or three years' time. And if an investor reviews a portfolio in a few years' time, it may be appropriate to rejig it, according to the latest forecasts.

Farrelly says this is different from "tactical" asset allocation, where a long-term, static allocation is "tilted" to take into account short-term imbalances in investment markets. Each time a portfolio is rebalanced or reconstructed, he says, it should be based on new, sound, long-term forecasts.

"The old approach is that you create a long-term static strategic asset allocation," Farrelly says.

"The idea of a long-term strategic asset allocation is that the relationships between different [asset classes] don't really change much over time [and] if that's the case, therefore it makes sense to have a stable asset allocation over time. The whole thing is built around efficient markets.

"That isn't a particularly difficult process to beat."

Farrelly says asset allocation and portfolio construction should be based on a few simple concepts:

■ Portfolios should be built to meet investors' cash flow needs;

■ Return forecasts should look ahead;

■ "Risk" is the chance of not meeting your cash flow needs;

■ There's more than one way to skin a cat (one portfolio does not suit all investors); and

■ Investors' preferences, transactions costs and taxes matter.