19th October 2010
Correlation is a slippery concept. In theory, it just means the extent to which two markets move in sync.
If, for example, the FTSE 100 and the S&P 500 both move up by 50% over six months, they would have a correlation of 1 over that period.
If bonds moved up by 25% over the same period, the correlation would be 0.5%. It bonds fell by 50%, the correlation would be -1. If they exhibited no relationship at all, the correlation would be 0.
However, the concept is more difficult than this bald analysis might suggest, largely because the relationship between two asset classes does not stay static. Therefore working out long-term correlations can be both complex and misleading.
Two asset classes that demonstrate no correlation in certain market conditions may suddenly become extremely correlated in a sell-off, for example.
This was particularly evident during the credit crunch when previously loosely-correlated assets such as private equity, hedge funds and corporate bonds, moved downwards in near-unison.
Why does this matter? It matters because a lot of portfolio construction theory is based round correlation. It assumes that historic correlation can be predictive of future correlation.
Harry Markowitz's portfolio theory states that combining non- or lowly-correlated asset classes can move a portfolio along the efficient frontier – i.e. enable investors to get the same or higher levels of returns with lower levels of risk.
For those interested in exploring more on Markowitz's theories the Financial Dictionary is not a bad place to start.
This is intuitive. If equities are falling, investors need an asset class that is doing well in their portfolio to mitigate those falls.
For example, it is an important reason behind the inclusion of alternative asset classes such as absolute return in a portfolio. Absolute return funds should show poor long-term correlation to equity and bond markets, thereby helping to smooth out difficult times.
The chances are that almost all portfolios will have been designed according to these principles. Therefore, if the theory does not work for any reason, it is bad news for investors.
This article from MSCI , and investment tools and research group, explains why the performance relationship between stocks and bonds is so important.
It is also becoming more of an issue. This FT article highlights how issues such as high-frequency trading and indexation are impacting correlation between equities.
It quotes research by Jeffrey Wurgler of the National Bureau of Economic Research, which shows that when a new company joins the S&P 500, its performance changes: "It begins to move more closely with its 499 new neighbours and less closely with the rest of the market.
"It is as if it has joined a new school of fish," he says, attributing this to indexation. His original research is shown here.
Seeking alpha is a good source of scholarly insight into correlation. It explores the correlation between different markets – and examines how investors can use correlation to build an effective portfolio.
The FT will show the beta of individual stocks in its markets section so investors can see how correlated an individual stock is to the index. Correlation between indices is harder to find, though comparative analysis is also possible on the FT site.
Correlation is a useful consideration when building a portfolio, but investors need to remember that it varies over time and today's non-correlated assets may not necessarily be tomorrow's.
It is proving particularly important at the moment, as analysts struggle to explain why it is increasing across equity markets.