Is diversification a bankrupt investment philosophy?

Since then, the benefits of diversification have become the accepted orthodoxy, and portfolio managers of all kinds have greatly increased their level of diversification. In recent years, many have added exposure to alternative investments such as hedge funds, property and commodities. But has the experience of the recent bear market exposed a flaw in the theory? The charge levelled by critics is that correlations increase when markets go down, making diversification not as effective a tool as had been assumed.

We disagree. Firstly, however, we should clarify exactly what we mean by diversification. It is perhaps easiest to understand at the level of selecting individual securities. A portfolio containing just one stock (Shell, say) is riskier than one containing two (Shell and BP). After all, each stock is open to all sorts of idiosyncratic risks – such as quality of management – which will not affect other stocks in the market, even if they are a competitor. So it makes intuitive sense that holding both Shell and BP would be less risky than holding either on its own.

However, Shell and BP are equally vulnerable to changes in the overall oil market. Both stocks would be likely to fall if there was, say, a sudden drop in demand for oil or a technological breakthrough in alternative energy. Hence, diversifying into different sectors, such as retailers, would reduce the risk of an unexpected event that affected the oil market.

The same principle obviously applies to diversification across regional markets and, even more importantly, to different types of assets, such as equities, bonds and alternative investments. But it also applies in areas that are less widely recognised. For instance, when selecting actively managed funds, we need to diversify across different managers, to reduce the risk from any one manager significantly underperforming.

But what is the evidence that assets have been falling in synch during the credit crisis? Looking at a graph of equity performance over the past 15 years, we can see that equity markets from the different regions of the world became more closely correlated last year – they were increasingly moving in step with one another. However, the graph also reveals that correlations have been rising steadily for many years. That should come as little surprise: the advance of globalisation is leading to closer links between world trade and world markets – and hence to higher correlations between markets.

Of course, one possible result of the current recession is that globalisation may suffer a reverse. If so, it should prove only temporary. Yet even a temporary reversal would reduce correlations and therefore increase the benefits of being diversified across different regional markets. And so should the fact that, after the credit crisis, lower levels of credit are likely to be available for years or even decades. It was the easy availability of credit that enabled hedge funds and other investors to borrow money to invest. When this credit was withdrawn, they were forced to sell securities to raise cash, driving down markets in tandem and thus driving up correlations.

Furthermore, even during the worst part of the bear market for equities, having a portfolio that was diversified across different asset classes – notably bonds and equities – generally delivered a significant boost to performance. In both the equity bear markets of this decade (see graph below), bonds have had a negative correlation with equities: when equity markets have been falling, bond markets have been gaining. So a significant holding of bonds will have helped to offset any losses from the equity component of a portfolio.

But merely ensuring that investments are not closely correlated may not be enough. Share A may be highly correlated with Share B in the short term, either rising or falling at the same time as one another on a daily basis. However, if Share B consistently rises slightly less than Share A and falls slightly more, Share A would significantly outperform over an extended period. Conversely, two investments may be completely uncorrelated – or even negatively correlated, when one falls if the other rises – and yet can still end up giving a similar total return. Hence, we have to look beyond simple measures of short-term correlation when building diversification into portfolios.

To conclude, diversification still brings benefits, both in theory and in practice. Indeed, the benefits may increase over the coming years. However, constructing a portfolio with effective diversification so that risks are genuinely lowered is no easy business. As you’ll see if you try reading Harry Markowitz’s original papers, the maths is a lot more complicated than the simple formula of not putting all your eggs in one basket.

This article expresses Coutts views as of 5 August 2009.

Pablo Balan, Global Head of Risk and Portfolio Analytics.