One of the oldest pieces of investment advice is that younger investors should hold more shares than older ones, because they have more years ahead of them in which to recoup losses; this is the logic behind 'lifestyle investing', in which people shift from equities to bonds as they get older. However, many economists think this view is wrong not just because bond prices might fall over coming years, but because the very idea is flawed in principle.
Let's start with the case for it. To reduce risk, we must hold assets that are uncorrelated with each other, so that if we lose in one, we have a chance of gaining on the other. But what assets have zero correlation with each other? One answer is: shares in one year and shares the next. Since 1870, the correlation of annual returns on the All-Share index in one year has been almost zero with those of the following year (minus 0.04 to be precise). If history's any guide, then, equity returns in 2013 will be uncorrelated with those in 2014, which in turn will be uncorrelated with those in 2015, and so on. If you hold shares in 2013, 2014 and 2015 you therefore hold three uncorrelated assets. That's safer than just holding shares in 2013. You can, in effect, use time to diversify. The longer your time horizons, the more uncorrelated assets you have, and the safer is your portfolio.
My table shows the point. It uses the actual average and volatility of annual returns since 1870 to estimate the probability of losing money in real terms. This probability falls with time. This is because if returns are uncorrelated from one year to the next, their volatility rises with the square root of returns, which is a slower increase than in the expected return. Shares, then, seem a safer bet in the long term than in the short.
The logic seems impeccable. But it's not.
Some of the problems with it were pointed out almost 50 years ago by the late Paul Samuelson, one of the founders of modern economics. There is, he said, a huge difference between adding risks and dividing them. When you diversify across assets - gold, equities, bonds, whatever - you are, albeit imperfectly, dividing risks. But when you diversify across time, you're adding them. You're taking your wealth from one period and betting it upon the next period. That's like making an accumulator bet.
This matters because although the probability of losing money in the long term is smaller than that of losing in the short term, it is a nastier event. A loss over 20 years could well destroy our hopes of retiring. A loss over one year doesn't. For many people, the small chance of a really nasty event is as unpleasant a prospect as the large chance of a less nasty one. For such people, long-term investing is no better than short-term investing.
There's a further problem. My table is based upon an heroic assumption - that future returns will resemble past ones. But there’s no guarantee of this. "Historical data tell us very little about the future," says Dennis McLeavey of the University of Rhode Island. "The stock market does not have a probability distribution." One reason for this is that although risk (known unknowns) declines with time, uncertainty (unknown unknowns) rises with it.
Will Goetzmann and Philippe Jorion, two US economists, have estimated that of the 24 national stock markets for which we have data in 1931, 10 subsequently suffered a long-term closure because of war, revolution or hyperinflation. How likely are such catastrophes in future? In the short term, we can be reasonably confident they won’t happen. But we cannot be so confident on a 20 or 30 year horizon - because there's so much more uncertainty.
Increasing uncertainty is especially important for investors with less diversified equity portfolios, because over longer horizons individual companies face uncertainty about whether bad management, competition or technical change will kill them off. Companies, remember, tend not to live long; of the 100 largest companies at the start of the 20th century, only three are independent quoted companies today. Short-term investors made nice profits in Enron, Marconi or Nokia - to name but three of many examples - but long-term investors did not.
There are two other complications here. One is that your time horizon might not be related to your age. For example, if you're planning on leaving bequests to your children or grandchildren, you have longer horizons than younger investors who are fending only for themselves.
A further complication is that younger investors, who might have longer horizons, have a big asset to consider - their human capital, or earning power. This affects their ability to take equity risk.
If your earning power is uncorrelated with the stock market - say, because you're likely to keep your job if the market falls - then you can afford to hold more shares because you can make up for equity losses by earning more and saving more.
However, over long periods, our human capital is likely to be correlated with equities because both depend upon the state of the economy; Japan’s 'lost decades' have seen not just big equity losses but also low pay growth and job insecurity. This correlation might argue for younger people investing in overseas equities - although this doesn't avoid the danger of a long-term global depression - but it's also a reason for them to hold fewer domestic shares.
To put this another way, older people - those with shorter horizons - might be able to hold more equities than younger ones because they face less risk that long-term losses on equities will be compounded by unemployment and worsening job prospects.
On balance, then, the conventional advice that equities are a good long-term investment is not as strong as it seems. Professor McLeavey says the merit of time diversification is "as elusive to prove as it is to disprove".
MORE FROM CHRIS DILLOW...
Chris blogs at http://stumblingandmumbling.typepad.com