Understanding Correlations
What correlation means for diversification and asset allocation
Correlation is a key statistical measure in investing. It is particularly important when it comes to diversification, asset allocation, and selecting suitable ETFs. It describes the relationship between two variables – for example, the performance of two asset classes – and allows us to determine how strongly and in which direction they change together.
In this article, we explore the basics of correlation, its significance for portfolio theory, and provide practical examples, such as those related to ETF selection. The main focus of the article is on correlation in the context of investing.
The essentials in brief: What Correlations Mean for Your Portfolio
Correlation: Correlation is a statistical measure. It indicates the strength of the relationship between two variables, such as the returns of two ETFs.
Correlation coefficient: The correlation coefficient is a numerical value between -1 and +1 that describes the strength and direction of the relationship. It applies to linear relationships.
Diversification: The lower the correlation among the assets in a portfolio, the more effective risk diversification is.
Tables and Visualizations: Correlation values can be clearly presented in tables or visualizations such as correlation matrices, dendrograms, or correlation tables.
Causality: A correlation does not prove a causal relationship, as spurious correlations may also exist. However, correlations can be an indication of a possible causal relationship, which should be critically examined.
Find and compare thousands of ETFs. Use powerful filters, detailed analytics, and independent data to discover the right ETF for your portfolio.
Definition: What is a correlation?
A correlation indicates the strength of the relationship between two variables within a statistical context. In this context, both variables either change in the same direction (positive correlation) or in opposite directions (negative correlation). Important: A correlation does not indicate a causal relationship (causality) nor does it prove one. It also does not specify which variable influences which. However, correlations can often provide clues about relationships.
The strength and direction of this relationship are indicated by the correlation coefficient, which can take values between –1 and +1.
What is a correlation coefficient?
The correlation coefficient “r” (also known as the correlation factor) developed by Bravais and Pearson is a statistical measure of linear relationships that indicates the strength and direction of the relationship between two numerical variables. Its value always lies between –1 (completely negative) and +1 (completely positive). A value of approximately ±0.6 or higher indicates a clearly discernible relationship, while a value of 0 indicates no relationship.
The Importance of Correlation in Investing
The goal of portfolio diversification is to minimise risk without sacrificing potential returns. This is achieved by spreading investments across different asset classes, sectors, regions, or currencies. However, diversification only works if the assets are not fully correlated with one another. Let’s take a closer look at the significance of the correlation coefficient “r.”
Positive vs. Negative Correlation
- Perfectly Positive Correlation (r = +1): The markets move in perfect sync with one another.
- Positive Correlation (r ≈ +1): The markets move almost in sync, for example, U.S. stocks and global stocks (r = 0.98).
- No Correlation (r = 0): The markets are independent of one another. There is no linear relationship.
- Perfectly Negative Correlation (r = -1): The markets always move in opposite directions and to the same extent relative to one another.
- Negative correlation (r < 0): The markets tend to move in opposite directions, as seen, for example, with gold and the global stock market (r = -0.04).
Join the extraETF community
Explore the world of ETFs and make informed investment decisions.
- Find ETFs that match your criteria with our ETF Search.
- Compare costs, performance, and risk side by side.
- Save your favourites to a watchlist and receive price alerts.
- Explore detailed ETF profiles with key metrics and holdings.
Analysing and Visualizing Correlations
In correlation analysis, time series are compared to identify linear or ordinal relationships. Correlation measures can be calculated using tools such as Excel, SPSS, or R. In financial analysis, correlation matrices, correlation tables, and dendrograms are particularly helpful for identifying groups of similar markets.
Correlation Matrix of Major Asset Classes
At the end of 2022, Stiftung Warentest examined the correlation among 67 global investment markets and asset classes. The analysis covered a 10-year period and used MSCI indices.
The result was a dendrogram. In this type of tree diagram, various investment markets are connected to one another via “branches” and “nodes.” The analysis also considered factors such as growth, value, and momentum, as well as themes, commodities, real estate, and more.
In addition to the dendrogram, this correlation matrix was created, which illustrates the correlation between the most important asset classes. Negative values are highlighted in bold.
For example: An investment in the global stock market (World) correlates with an investment in emerging markets (EM) at a value of 0.66. This means that if global stocks rise, emerging markets tend to rise as well – and vice versa. However, the correlation only describes the strength of the statistical relationship, not the exact percentage change.
| Global | U.S. | Europe | Japan | DAX | EM | Gold | Euro-denominated government bonds | |
|---|---|---|---|---|---|---|---|---|
| World | 1 | 0.98 | 0.90 | 0.76 | 0.84 | 0.66 | -0.04 | 0.34 |
| U.S. | 0.98 | 1 | 0.81 | 0.70 | 0.76 | 0.59 | -0.04 | 0.34 |
| Europe | 0.90 | 0.81 | 1 | 0.69 | 0.95 | 0.65 | -0.06 | 0.30 |
| Japan | 0.76 | 0.70 | 0.69 | 1 | 0.68 | 0.59 | -0.13 | 0.26 |
| DAX | 0.84 | 0.76 | 0.95 | 0.68 | 1 | 0.63 | -0.10 | 0.25 |
| EM | 0.66 | 0.59 | 0.65 | 0.59 | 0.63 | 1 | 0.08 | 0.26 |
| Gold | -0.04 | -0.04 | -0.06 | -0.13 | -0.10 | 0.08 | 1 | 0.22 |
| € Government Bonds | 0.34 | 0.34 | 0.30 | 0.26 | 0.25 | 0.26 | 0.22 | 1 |
Key Findings from the Correlation Matrix
The most important correlations and relationships between the markets can be derived from the correlation matrix and the dendrogram as follows.
Very high correlations
- Global and U.S. 0.98: This very high correlation is not surprising, given that one criticism of the MSCI World Index is that it consists of more than 70% U.S. stocks. This poses the risk of concentration risk.
- Europe and the World 0.90: Here, France, Germany, and the United Kingdom form a cluster. The European average is 0.97, a figure France comes closest to.
- Dividends and Value 0.93: These trends are quite similar.
- Canada and commodities 0.84: This could be attributed to Canada’s abundance of natural resources and its strong industrial presence in the sector.
- Switzerland and the international healthcare sector 0.82: This is also supported by the fact that over 35% of the companies in the MSCI Switzerland index are from the healthcare sector.
- Norway and the energy sector 0.79: A correlation with the country’s energy resources, particularly in the field of renewable energy, is likely.
- China and Hong Kong 0.79: Although China is classified as an emerging market and Hong Kong is considered a developed country, the two are closely linked historically and culturally. Numerous funds invest in both countries.
Low correlations with the MSCI World
- Individual emerging markets, e.g., 0.28–0.32: Emerging markets such as Turkey (0.28), Egypt (0.31), and China (0.32) exhibit lower correlations than emerging markets in general (0.66).
- Euro-denominated bonds, e.g., 0.34–0.63: government bonds (0.34), inflation-protected government bonds (0.58), and corporate bonds (0.63).
- Industrial sectors, e.g., 0.58–0.74: Utilities (0.58), Energy (0.66), and Consumer Staples (0.74).
- Individual developed countries, e.g., 0.48–0.58: Hong Kong (0.48), Portugal (0.5), and New Zealand (0.58).
Largely uncorrelated
- Gold exhibits almost exclusively negative correlation coefficients (-0.04) in the correlation matrix with the MSCI World. Consequently, its node in the dendrogram is also close to zero.
The complete correlation analysis can be found on the Stiftung Warentest website.
ETF Diversification and Correlation: A Powerful Duo
Although ETFs are, by definition, broadly diversified, they are not automatically well diversified. A portfolio consisting of multiple ETFs offers additional risk diversification only if the ETFs are weakly correlated with one another and there are as few cluster risks as possible.
One example is the comparison between the MSCI World and the Nasdaq 100:
The indices tracked by these ETFs contain many U.S. tech stocks, such as those from the “Magnificent 7” – companies like Nvidia or Microsoft, which are included in both ETFs and carry heavy weightings. Combining these ETFs quickly creates a concentration risk. A better result can be achieved by adding an emerging markets ETF (e.g., MSCI Emerging Markets) or a gold ETC with a significantly lower correlation coefficient.
Limitations of Correlation: What Investors Should Know
Correlation is a powerful and insightful tool, but it has its limitations. Keep the following in mind:
- Correlation does not imply causation: Just because two data series are related doesn’t mean that one is the cause of the other (keyword: spurious correlation).
- Temporal instability: During times of crisis, correlations often spike, which makes diversification less effective.
- Impact of outliers: Extreme values can skew the correlation coefficient.
- Reducing portfolio risk: Combining markets and asset classes with low correlation can help improve your portfolio’s overall risk profile. Make sure to maintain a balanced asset allocation.
- Volatility and Total Losses: Despite correlation analyses, investments in emerging markets in particular can be highly volatile. Total losses, such as those seen in the Russian stock market, are always possible and cannot be predicted by correlation analyses.
- Correlation Does Not Equal Return: Just because markets are correlated does not automatically mean they will generate the same return. The reason: Highly correlated markets tend to move up and down in the same direction. Nevertheless, they can fluctuate to varying degrees.
- No Guarantee: Making investment decisions based solely on correlation analyses is risky. However, they can give you an idea and a sense of which securities and asset classes are typically strongly linked.
Conclusion: Leveraging Correlations Strategically for the ETF Strategy
Correlations are an essential tool for identifying and taking into account relationships and risks related to portfolio composition, investment strategy, and asset allocation before making an investment. Correlation matrices or dendrograms can provide valuable insights into correlations and correlation coefficients.
More Questions About Correlation and Investing
- What Correlations Mean for Your Portfolio
- Definition: What is a correlation?
- What is a correlation coefficient?
- The Importance of Correlation in Investing
- Analysing and Visualizing Correlations
- Correlation Matrix of Major Asset Classes
- Key Findings from the Correlation Matrix
- ETF Diversification and Correlation: A Powerful Duo
- Limitations of Correlation: What Investors Should Know
- Leveraging Correlations Strategically for the ETF Strategy
- More Questions About Correlation and Investing
