BNP Paribas uses cookies on this website. By continuing to use our website you accept the use of these cookies. Please see our cookies policy for more information and to learn how to block cookies from your computer. Blocking cookies may mean you experience reduced functionality or be prevented from using the website completely.

#Investments — 29.04.2016

Correlation One Day Doesn't Mean Correlation Forever!

Sophie de La Chapelle

The correlation or strength of the relationship between two assets is often believed to be permanent. This isn't the case at all...

It is very important to diversify when creating the best portfolio structure to allocate assets. This means, that three key parameters must be considered: the risk/return profile of each asset class, the sensitivity of each to economic factors (to growth and inflation in particular) and, especially, their mutual correlation. Correlation analysis is indeed a critical step in building a portfolio. More specifically investors must be aware that correlations fluctuate over time: because, if they believe that correlations are stable over time, they run the risk of underestimating portfolio risk and overestimating the benefits of portfolio diversification.

That is the reason why I think it is essential to focus on this statistic and on its fundamentals in order to be not only cautious when using correlations but also lucid about the potential limits of the exercise.

But what does correlation mean exactly? What are the keys to understanding this important concept?

When you hear professionals talk about correlation, they're often referring to stocks and bonds. The main reason for this is that the way in which wealth is allocated between these two asset classes is considered as a key aspect of diversification. Stocks correspond to securities of listed companies whose price varies from second to second. They are perfect assets for investors willing to take on greater risk hoping to enjoy greater gains. On the contrary, bonds correspond to debt issued by companies, institutions and governments. The price of bonds can also vary, but it is generally less volatile than that of stocks. Stocks and bonds tend to have an inverse relationship. Shares usually go up in value when economic growth accelerates. On the other hand, investors invest in bonds or bond funds when they're anxious about the future and want to protect themselves.

What is its role in the portfolio?

This means that a mix of shares and bonds guarantees a diversified portfolio because, when shares go up in value, bonds tend to go down, and vice-versa. In other words, stocks and bonds are negatively correlated. Therefore, this is a way to protect your portfolio against the fluctuations of one of the two asset classes and optimise the level of risk of your portfolio.

Does it change over time?

Let's consider the nature and characteristics of this correlation for a moment. Correlation is a statistical value, calculated based on past data (profitability) and, contrary to what you might think, it isn't fixed at all and can vary over time. Here, it means that stock and bond markets have historically diverged over time. However, this long-term trend can fluctuate over the short run under certain market conditions. For example, let's look at two specific time intervals: the 2007-2009 financial crisis and the quantitative easing period in the United States during which the American Central Bank injected massive liquidity into the marketplace.

  • The subprime crisis of 2007-2009 was a typical example of a change in the correlation between different asset classes. Following the bankruptcy of Lehman Brothers, investors were compelled to liquidate their asset portfolios, selling them at any price, and leading to a re-correlation of different asset classes. Investors had very few, if any, places to take refuge other than in cash (liquidity). The long-term diversifying properties of bonds compared to shares temporarily disappeared.
  • The American central bank's more recent quantitative easing also provides a very good example of a period during which correlations were temporarily turned upside down. The American Central Bank contributed both to the lowering of bond interest rates, which increased the price of bonds, and to the rise in the stock markets (a search for higher yields than on monetary and bond investments). Stocks and bonds were temporarily positively re-correlated.


At the end of the day, the correlation between shares and bonds depends on macroeconomic factors including inflation, growth, the unemployment rate and the level of real rates.

Of course, this isn't the only example of a change in the relationship between two asset classes over time. Let's consider oil and European stocks. As can be seen in the graph below, the correlation between oil and European stocks tends to be positive (that is, when the price of oil goes up, European stocks also tend to go up). However, it's quite clear that the strength of this correlation varies over time. During the first half of 2015, the drop in oil prices was seen as a positive factor (i.e. lower costs for companies and higher consumption) and stocks tended to go up in value. The correlation between the two assets was not very strong, or even negative. This relationship has changed since the end of 2015. Shares go up when oil goes up and shares fall when the barrel price drops. What's the reason for this? First of all, a drop in the price of oil has a negative impact on the economic prospects of producing countries. This means that Brazil and Russia are in recession. Moreover, the energy sector accounts for a significant portion of stock markets (the "majors" like Total and Royal Dutch Shell, and oil services companies). Lastly, the drop in the price of oil makes costly drilling unprofitable and upsets the business model of certain companies, notably American shale oil companies. Mixed with increased concern for growth prospects, this has logically contributed to draw stock markets downward since the beginning of the year.

Preserving diversification benefits

Analysing correlations is an important step in building a portfolio. Based on the preceding examples, it's clear that the relationship between asset classes, which are sometimes strongly anchored in the minds of investors (some more than others), are not stable over time. The most radical times of instability are obviously those during which the markets are under stress and upset these relationships. At these times we talk about a "correlation breakdown" (from an empirical standpoint, we can see that correlations in extreme situations are often very different from those observed under normal conditions and that we tend to underestimate the probability of "joint" negative returns during these periods). Investors must be aware of these fluctuations and should be able to adjust their portfolio allocation if necessary, in order to ensure an efficient diversification and manage risk appropriately.

Correlation Between Oil (Brant) And European Equity

In order to measure correlation, we calculate the correlation coefficient that is always between -1 and 1. Thus, two variables with a correlation coefficient close to 1 will trend in a very similar way. If, however, such coefficient is close to -1, they will evolve in opposite directions. Finally, if this coefficient is close to 0, the two variables are considered as independent and said to be uncorrelated. Source: Bloomberg, correlation between oil (Brent) and European stocks (Eurostoxx 50) over 120