Short-term forecasts and other silly questions

In December each year, one of the fund managers has a Christmas lunch where, amongst other activities, the winner and loser of its annual ‘market predictions’ competition is announced.

Predictions for the All Ordinaries, the S&P 500, gold, oil, the Australian dollar and Australian 10-year bonds are made one year in advance. Prizes are awarded to the person with the best and worst forecasts based on the lowest and highest average dispersion of the six predictions from the then current figures.

For 2002 I won this award. Almost spot on the S&P 500, All Ordinaries and the gold price, I was off a modest amount on the Aussie dollar and 10-year bonds, and a fair way off the volatile oil price.

Now before you ask where you can subscribe to my predictions for this year, I must firstly disclose that I received previous awards for coming last in this same competition two years previously and second last in the previous year (the person who came last was obviously too embarrassed to turn up).

Of course, my defence is that my predictions from the previous two years were not wrong — it just took three years for them to come through. In any case, while this game was never meant to be serious, it got me thinking about the value (if any) of short-term forecasts and predictions and the way in which many in the investment industry approach making (and receiving) them.

At the beginning of every year there is always a flurry of predictions and forecasts for the year ahead. Like horoscopes (and perhaps as useful), it seems many people have an addiction to both seeking out and making such forecasts. Do they really have any value though?

My view is that predictions of most financial variables on a one-year time frame are, by themselves, mostly useless, and perhaps quite dangerous if taken at face value. I have long held that one of the silliest questions regularly asked of fund managers and strategists at presentations is, “Where will the All Ordinaries index be in a year’s time?”

This is a silly question because:

Consistent, accurate forecasts of most financial variables on a short-term basis (and one year is short-term) are impossible; and

One-year forecasts are not that relevant to the average investor anyway.

In answering this question most ‘experts’ don’t really even attempt to forecast — they simply provide a near consensus figure that has little if any value.

The major problem with short-term forecasts is there is too much random movement in response to unexpected news and sentiment over such a time frame. For example, the range in which equity indices move over a year can easily be 20-30 per cent. Even if right on the medium to long-term trend, such noise can easily blow out any one-year prediction.

One also has to ask how relevant one-year forecasts of markets are in any case. For example, most investors don’t have a one-year investment horizon. Nor are they just invested in an index. Further, because most investors are accumulating or drawing down assets over time, predictions for one point in the short-term future are of little use.

This difficulty and lack of value from short-term forecasting leads some commentators and advisers to jump to the conclusion that all active investment management should be abandoned in favour of passive alternatives. This is a mistake. The difficulty and irrelevance of one year forecasts has very little to do with market efficiency or the ability of managers to earn above market returns relative to risk taken.

For example, can you reasonably expect that high market valuations compared to history mean lower long-term return going forward? Can you reasonably judge that sectors that have been out of favour in terms of valuation, funds flow and sentiment have a better chance of generating higher long-term returns than those sectors that are in favour, expensive and receiving lots of hot money?

Can you find good stock pickers capable of finding undervalued companies likely to outperform the market? Can you expect that buying a listed investment company at a 25 per cent discount is more likely to produce better results then one trading at a 30 per cent premium?

None of these involve predicting where the market will be in 12 months time. But all of them can add value to an investment portfolio.

In many ways it is easier to make useful forecasts of returns over 5-10 years than it is over one year. This is because the noise that affects a one-year forecast is diluted over this time and the behaviour of the variables that drive stock prices are reasonably consistent in the long-term. Over the longer term, earnings tend to grow at a reasonably constant rate, and valuations tend to revert over time. Such simple analysis was screaming for flat or negative five to 10 year returns on the US market from the late 1990s. Of course, few were listening, too interested in where the market was going in the next year.

It is also interesting to see how the investment industry approaches making these short-term forecasts. Perhaps because they realise how difficult it is, most tend to take a very safe route. Both in the fund manager game and in the real world, the most common forecasting strategy is to take the current level and either use that, use five to 10 per cent or simply extrapolate forward one year, the long-term average.

This is also what you typically hear from the average economist, stockbroker or fund manager at the beginning of the year forecasting parade. Such forecasts tell you nothing, except that the forecaster wants to keep their job and they believe that listeners are gullible enough to take their forecast seriously. The better answer in many cases that you hear all too infrequently is, “I don’t know”.

Where a short-term forecast might have value (although not by itself) is when it is significantly different to the general consensus, and usually indicative of a longer term view. In this case, the key is not the figure or return itself but the supporting rationale as to why a market should go up or down. After all, if you have sound reasons to avoid a market and it falls 15 per cent instead of the 30 per cent predicted, the forecast is still very wrong. But you have avoided losing money — why should this be seen as a mistake?

Likewise, if a market goes up 30 per cent instead of the 15 per cent forecast and the view was backed by being invested, how relevant is it that this forecast was still wrong?

A forecast away from the consensus is at least trying to tell you something. It may be naïve to attempt this on a one year view, as the random swing of markets may render it wildly inaccurate, but this does not mean that the premise for the underlying view is necessarily wrong. You cannot outperform in investments just by knowing and doing what everyone else is doing. In this case, the actual forecast number is not really what’s important.

In the end, managing investments is not about short-term prediction. It is about judging risk and reward and assessing scenarios and probabilities. It is about finding undervalued and selling overvalued securities. It is about having a feel for market history and market cycles and where we are in that cycle. The difference between these activities and one year market predictions is night and day.

However, if ever you are in a short-term prediction game where you can win prizes for coming both first and last, it makes sense to at least have a go.

Oh, and my answer is that the Australian All Ordinaries index will be 2,700 at year-end. But again, what would I (or anyone else) know? And it’s still a silly question!