Portfolio optimization with increasing correlations in our evolving world

Over the years, investors have viewed diversification as the only “true free lunch.” In fact, asset classes such as global stock markets, real estate, lumber, commodities, managed futures, and other alternative assets have served their advocates well. On the other hand, some analysts argue that the benefits of diversification unravel at the worst possible times. This appears to be true, as evidenced by the recent financial crisis, in which well-diversified portfolios declined by -25% or more. How can both sides of the argument be true? Is there anything an institutional or endowment investor can do?

Using best practice institutional investment and hedge fund strategies, and applying a mathematical and scientific approach to improve statistical and risk management concepts, can maximize the use of information and diversification potential available. It is useful to apply theoretical approaches sensibly to ensure practical and robust results in our pragmatic world. The result is a more complete model that combines Monte Carlo, Post-MPT analysis, and more meaningful risk measures. Below are some ideas about these statistical measures and methods.

Global Actions, Increasing Correlations, and Semi-Correlation

Beginning in the 1980s, international stocks were the hot investment category. They added diversification to a well-diversified portfolio. The Japanese stock market went from around 10,000 to around 40,000 during the 1980s and helped stimulate interest in foreign stocks. The US, European, and Asian stock markets have always been correlated with each other, but correlations were typically in the 0.4 to 0.7 range prior to the mid-1990s.

Mean variance and other Modern Portfolio Theory models were “happy” to see these relatively low correlations. Portfolio optimizers have shown that you could slightly increase your equity exposure overall, allocate a significant amount of your equity exposure to other regions of the world, and still increase the overall risk/return characteristics of your portfolio. Over the years, international stocks (rather than just home country stocks) have served diversified portfolios well.

However, as with most good ideas, the benefit of international stocks has diminished over the years. Mathematically, there will always be some benefit to global stocks, but the numbers show generally increasing (moving) levels of correlation over the years. Correlations between foreign stocks and the S&P have risen from an average of around 0.5 or 0.6 in the late 1980s and early 1990s (when international stocks began to catch on) to current levels of around 0.8 or 0.9.

Key takeaways:

  • Correlations between world stock markets have generally increased over the years; the benefits of diversification diminished.
  • Interestingly, there are spikes in the correlation, especially in times of financial crisis. Consider the Crash peak of 1987, as well as the very high correlations during the current recession.
  • The previous point quantifies the observation of many investment analysts: that the benefits of diversification in many asset classes are less than expected.


In general, we have seen markets sometimes fall together, and the benefits of diversification dissipate, at the worst times. When there is turbulence, markets become more correlated as portfolio managers cut losses and try to maintain liquidity. I have developed proprietary indicators (* is an example, below) to determine if diversification could really help in times of need.

Correlations and semi-correlations for the S&P 500 and various sectors (1987-present)

Nasdaq-S&P Correlation = 0.84
Europe-S&P correlation = 0.80
Asia-S&P Correlation = 0.69
Nasdaq-S&P = 0.95

Europe-S&P = 0.93


Asia-S&P = 0.82

I sometimes mention “half-deviation” as a better measure of overall risk than standard deviation (because it measures downside risk). Semi-correlation is a similar approach that removes some of the noise (noise due to bullish moves/correlation) and tries to measure “times of trouble” more directly. In the chart above, we can see that correlations increase during financial market volatility. More specifically, the chart shows that when the S&P fell, the Nasdaq, European and Asian markets were down about 90% of the time. In fact, if we study the “material” falls, the diversification figures worsen until they are close to 100%.

Real estate correlation over time (1982-present)

Real estate is another asset class that has provided good diversification over the years, with a long-term correlation to stocks of around 0.1. Based on data from 1982 to the present, we have seen correlations increase from near 0.0 to recent correlations near 0.3 or higher, and the recent financial crisis is closely related to the real estate sector. Summary

The correlation of some asset classes has increased over the years. Furthermore, history has shown that the actual benefits of diversification are lower than expected, as markets fall together during market downturns. Using a good set of tools can help investors gain a more realistic understanding of the odds. These tools have uncovered some interesting relationships between asset classes and strategies. Furthermore, financial markets and the world around us are constantly evolving. The value of a good idea often diminishes over time. What will be the next big investment idea? It is important to constantly improve to remain competitive in our ever-changing world. Continuous research and a collaborative effort, with a skilled investment team, can help an organization stay ahead of the crowd and achieve good risk-adjusted returns.

Leave a Reply

Your email address will not be published. Required fields are marked *