Buying shares and holding them for the long run might make intuitive sense but unfortunately it doesn't guarantee a happy ending.

If you happen to buy shares when they are highly valued you are likely to be disappointed. March 2000 was a time of significant overvaluation in the US stockmarket.

At that time the S&P500, a measure of the American stockmarket, was at 1359.15. Some 12 years later the index is virtually unchanged and this measure excludes tax and fees. The latter could reduce returns by as much as a further 2 per cent pa.

So contrary to the "buy and hold, invest for the long term and you will be okay" fairytale, keeping an eye on valuations is critical. Research by Clifford Asness cited here in August 2005 highlighted the fact that buying shares at high valuations consistently led to poor long term returns. This is hugely significant because valuation movements can dominate the sharemarkets natural gradual upward movement due to growth in earnings.


Profits grow by something less than what the economy grows by ie 3-5 per cent pa, but valuation movements can be much greater than this. According to Longview Economics, a popular valuation method known as the cyclically adjusted PE (CAPE) varied in the period 1900 - 2011 from a low of 4.8x to a high of 44x. So the stockmarket at peak valuation was valued 10x higher than at its lowest valuation.

But buying low is easier said than done.

There is of course no shortage of information on investing in the sharemarket for mum and dad contemplating equities as a home for their savings. Trouble is that much of it, for one reason or another, is misleading.

Take past performance as an example: many people think that it's the bottom line as far as picking what's going to do well in the future. However most times this is dumb because when shares have performed well in the past they usually underperform in the future because their price is high and their dividends are low. There are a number of reasons why discerning whether prices are high or low is a difficult thing to do.

Firstly there is the problem of delayed feedback. If you drive on the wrong side of the road you will get a reasonably immediate signal that this isn't a sensible thing to do when you hit another car. But on the stockmarket you can make lots of money for a long time despite investing at what will later be seen as being obviously overvalued levels. The stockmarket can stay at overvalued levels for years - take 1999 to early 2008 for example.

Historical performance figures as at January 2008 for a hypothetical leveraged hedge fund with a simple strategy of just borrowing half its money and investing in the US stockmarket show the manager to be a genius with a return of about 19 per cent pa for 5 years. Eighteen months later and he or she is a fool who has lost his/her clients about three quarters of their money.

Getting a "true and fair" view of the stockmarket is further compounded by the fact that stockbrokers, financial planners, fund managers and investment bankers need cash flow 100 per cent of the time but, according to Andrew Smithers, independent economist based in London, the stockmarket has of course has only been undervalued approximately 50 per cent of the time in the period 1900 to 2011. In contrast buy signals in some form or another are required 24/7.

Despite these impediments, as Mr Asness illustrated, it is possible to infer some useful view of valuation from long term sharemarket data. But this means there will be times when you have to tell clients that stocks are too high, a big call and definitely not good for business in the short term.


Smithers research prefers two methods of valuing stockmarkets and under both measures he concludes that the US stockmarket is currently 50 per cent overpriced. The two valuation techniques are "Q" and the cyclically adjusted PE (CAPE). We will focus on CAPE here as it is the easiest to understand.

The conventional price earnings ratio of a stock is just its price per share divided by its earnings per share. So if stock A trades at $6.00 per share on the stockmarket and its profits are $1.00 per share it has a PE of 6. Easy. CAPE is the same deal but it uses the average of historic earnings over the last 10 years to get the earnings per share figure and thereby gets around the cyclicality of profits.

Some years profits are high, some years they are low due to, amongst other things, the state of the economy. Mr Smithers points out for example that the earnings per share of the US stockmarket for the 12 months ended 30 June 2009 was $7.51 so that the PE, with the index at 1.073, was 143 which is 10 times the long term average using data starting in 1871.

Mr Smithers details various methods whereby analysts manage to manufacture buy recommendations based on faulty analysis and he gives as examples the tendency to compare price earning ratios based on future earnings with average PE's based on trailing EPS. As the past trend in earnings has been upwards this tends to make the market look cheap. Secondly analysts can use operating earnings rather than actual earnings. Operating earnings excludes extraordinary write-offs but obviously in the long run extraordinary write-offs occur ordinarily.

So what practical use is this? Obviously if one believes it one should avoid US stocks but with correlations as they are it is a good bet that if the US goes down so will everything else. For the practitioner who needs to generate cash consistently to stay in business an application of this research might be to rebalance into more bonds.

But 10 year bonds in the US yield just 2.2 per cent and are arguably more overvalued than shares.

The answer, having regard to the fact that the stockmarket can stay overvalued for a long period of time and the need for stockbrokers and financial advisers to stay in business is probably to take notice of valuations but instead of selling out of equities completely to reweight into asset classes which look less expensive and today that probably means having less money in US shares and bonds and more exposure to the local market.