Dominic Wallington: When is a bubble not a bubble?

The late comedian Peter Cook once pointed out that the Great Train Robbery "in fact involved no loss of train, merely the contents thereof. Trains are somewhat difficult to lose, given their size." Using the same logic, I would like to point out that the TMT bubble in fact involved no bubbles whatsoever. This may appear a somewhat obvious point, but I make it in order to attack the imagery rather than to be pedantic. Burst bubbles disappear immediately and have little residual impact on their surroundings.

To understand how stock markets work, it is better to think of the events of the last two years as the TMT balloon. This was untied to travel noisily and incautiously across a large room, only to come to rest when all of its excess had been released. It is easy to see why this image didn't catch on and the bubble idea did.

Ultimately, though, the point being made is a serious one. The sorts of returns that are available after a period of over-investment may not be immediately apparent to all those involved. This dislocation can create very volatile markets. The fulcrum here is human behaviour, something dismissed by the initiators of Modern Portfolio Theory.

Nietzsche said that most individuals are sane while most crowds are not. Crowds regularly behave in a way that the constituents, as individuals, would never countenance. Crowd behaviour is both magnificent and complex, and it is astounding that anyone believes this type of behaviour should not affect a market (which is another collective).

Too much capital applied in any one industry will bring down expected growth, and ultimately every penny spent in capital can see a declining return. The higher the returns initially, the more rapidly they tend to move to the mean. The problem is that most investors and company managers are conservative: they make up their minds and don't like to change them. So if they have decided that very high growth and returns will always be available, it will take some time for them to change their view. This conservatism can be demonstrated by the fact that technology earnings growth peaked about six months before share prices.

Let us examine the trajectory of the balloon during a boom period. In the early correction stages, share prices tend to decline sharply before stabilising. People remain optimistic that the favourable growth differential will be re-established after a period of consolidation. But after more months of earnings deceleration, share prices collapse.

It is difficult to know how far through this process we are, but after 18 months of under-performance, investors are now buying the afflicted sectors. As we have seen, however, it is necessary to caution against the assumption that a sustained recovery is under way just because there are companies and investors who believe it is. Not only do supply-side issues such as cost-cutting and restructuring need to be addressed, but demand must also return.

As Peter Cook observed, things are not always what they seem. Companies and investors can follow general rules about valuation and returns, but there is sometimes a fear that a new world order has arrived and the rules have changed. In such a situation my advice would be to stick with a core of successful companies that have earned above their cost of capital over at least one cycle and have high barriers to entry. If expected returns across the market or in individual companies rise fundamentally above long-term averages, detective work needs to be undertaken to establish whether they are sustainable.

Dominic Wallington is UK fund manager at Invesco.