• X
• Facebook
• LinkedIn
• Reddit
• Mastodon
• Pocket
• Mail
• Print

Does it really work?

The short answer is — yes, but there are important caveats. Portfolio diversification is a core concept in finance and there is almost no financial adviser or planner who would not encourage their clients to consider it as a prudent risk-reduction technique. At least to the extent it is reasonable and suitable, of course. But behind the sound theory illustrating its apparent benefits, diversification is in practice a more complicated phenomenon. It can vanish when you need it the most, and it can resurface when you want it the least. Knowing when it does or does not work can make you a much savvier investor, better prepared to handle various market environments.

The content of this post does not constitute financial or investment advice. It is intended for information and education purposes only.

You can also read this article on Medium

How is it supposed to work, at least in theory?

The idea of diversification is rather simple — if you spread the risk exposure of your portfolio across several, little or not related assets/asset classes or factors/drivers of risk, then you should benefit from the fact that they perform differently in various economic and market environments. If we think about typical investments and look at their historical returns, we would notice that, over time, the returns have been highly variable for some of them, while for others much less so. For example, rolling 1-year US stock market returns have ranged from -37% to over +50% between 1872 and 2023, while bonds have returned on the same basis between -13% and +43%. And their behavior over time is not perfectly aligned.

Nobel laureate Harry Markowitz developed the so-called Modern Portfolio Theory (MPT) in the 1950s as he studied this phenomenon, which suggests that risk-averse investors should attempt to construct portfolios of multiple assets (or asset classes) in a way that gives them the return they are targeting with the least possible risk (or, obtain a maximum possible return for the level of risk that is acceptable). He established a mathematical framework that shows how putting together multiple risky (stock) positions could be less risky than each individual stock in it.

This discovery highlighted the importance of correlation — i.e., the measure of co-movement between things, in this case — investments. Correlation is usually measured as a coefficient, ranging from -1 (i.e., perfect negative correlation — meaning, the investments would always move in perfect tandem opposite to each other to the same degree) to 1 (i.e., perfect positive correlation — moving to the same degree in the same direction, up or down). If a correlation coefficient is zero, then the assets are said not to be correlated. The implication is that you can avoid sharp fluctuations in the returns and value of your portfolio by owning assets with correlations that are not perfectly positive.

Myth #1: You are sufficiently diversified with 20–30 stocks

Studies of the markets have shown that simply increasing the number of securities in a portfolio has its limitations. Beyond a certain number, the additional benefits diminish in terms of reduction in portfolio volatility. Some research has found that the “magic” number is 20–30, while others found numbers like 60–80. However, if criteria other than simple variance (or standard deviation) are used to determine this, the number could actually be much higher. For instance, a recent paper demonstrated that when using the terminal wealth multiple, the distribution of annual volatilities, and the distribution of annualized returns, a long-term investor may require over 250–500 stocks or more. So, no fixed number would solve this puzzle, but there can certainly be additional benefits of diversifying beyond the usually assumed 20–30 stocks.

Myth #2: ‘Set it and forget it’ ensures you are diversified forever

Unfortunately, correlations are not constant — they change over time depending on economic and financial conditions, monetary and fiscal regimes, etc. Correlations are not always symmetric on the upside and the downside. Additionally, as the value of your investments evolves over time, the exposures to certain risk factors would also change, meaning that you might still end up more exposed to risks than you are not able or willing to tolerate.

If we look at the US stock and bond markets, for instance, research shows us that over the last 20 years, bonds managed to not only be a primary diversifier for equity portfolios but also to deliver positive returns in periods when equity markets suffered losses. Novice investors who have not been investing for many decades basically had that as their entire experience of how this relationship works. However, that correlation hit a 40-year high during 2023 as investors adjusted their expectations to the new ‘higher-for-longer’ monetary regime in an environment of high inflation. Indeed, if we look back in history over 100 years, we will see that negative stock-bond correlations have been the exception, rather than the rule.

The drivers of the variation of the stock-bond correlation could be multiple, but the primary ones from a macro perspective are growth and inflation. According to this study, stocks and bonds have opposite-sign sensitivities to news about growth and same-sign sensitivities to inflation news. For example, if growth is higher than expected, investors would anticipate future cash flows to increase, leading them to bid up the prices of equities. However, they would also anticipate that short-term rates might start increasing due to the rise in the equilibrium real rate and the reaction of the central bank which would want to prevent overheating of the economy and inflation. Hence, bond prices would fall. Other factors impacting the correlation are for instance monetary policy shocks, credit risk, or even sectors (stocks from certain industries could be more correlated with bonds than others) and styles (e.g., growth vs. value).

In an ideal world, rational investors would prefer to have a situation where correlations are low on the downside (i.e., when for instance equity markets are crashing — so-called “left-tail” events) and high and positive on the upside (“right-tail”) — that is, unification instead of diversification. That way, you benefit from diversification only when you really need to be protected from significant loss and get gains from as many assets as possible when they move favorably for you. But that is not how the real world works. This requires looking beyond “full-sample” correlations which can be very misleading.

Conditional correlations that focus only on the left or right tail show that:
– multiple types of risk assets are highly correlated with US equities when they are crashing, while those correlations are much lower (or even negative) when the opposite happens;
– similar observation when comparing hedge fund styles to US stocks;
– private assets such as private equity or direct real estate (when smoothing bias is removed) do not offer useful diversification benefits for equities in times of market stress;
– stock markets in G5 countries are 2 times more likely to co-crash compared to bond markets;
– currencies co-crash more often than expected if using a bivariate normal distribution.

With factor exposures, the picture is a bit mixed. The definition of risk factors usually involves designing long-short portfolios of securities, which is where the strength of some factor strategies lie and they might appear to offer better diversification. If the long-short constraint is removed for the overall investment universe tested, then factors such as size and currency carry could fail to provide diversification when needed. Momentum strategies, however, that involve selling risk assets in down markets, provide more robust left-tail diversification.

Myth #3: Domestic diversification suffices

This typically depends on the investor and their goals and time horizon, risk tolerance and constraints. Generally speaking, if you are investing for the long run (say, 20 or more years), chances are you will experience periods where the domestic market would be underperforming relative to international markets. Take the below example — US vs. non-US stock returns over a near-50-year period. You can clearly see how outperformance alternates and US equities do not always do better than their global counterparts.

However, even this idea of “international diversification” has its caveats. The relative benefits can depend, for instance, on whether an investor is restricted to investing only in developed markets (DM) or not. Research shows that if there is such a restriction, after the year 2000, it is generally better to aim for cross-industry diversification rather than cross-country diversification. For example, if we think about Western Europe and North America, there is significant synchronicity in how equity markets move (also supported by the fact that globalization has contributed to the much closer integration of these economies and their capital markets). In contrast, if there is no restriction to invest in emerging markets (EM), or if only EMs are allowed, cross-country diversification could be a better choice.

Myth #4: The benefits of diversification disappear during crises and market crashes

Since recessions, debt crises, market crashes, etc. can significantly influence capital flows, they can also impact diversification benefits. For instance, the very notion of diversification came under attack around the Global Financial Crisis (2007–2009) when it seemed to fail as correlations spiked across many major asset classes and markets abruptly switched to a risk-off mode. But factor diversification has remained generally consistent even when investors have panicked. And some studies of periods of significant turmoil (such as the Asian Financial Crisis 1997–1998, the Dot-com bubble burst 2000–2002, the GFC 2007–2009, the European debt crisis 2009–2010, the post-Brexit uncertainty 2016–2017 and the COVID pandemic 2020–2021) demonstrate that the benefits of international and industrial diversification can actually be robust and countercyclical.

But these do not remain equally significant over various periods. In fact, this paper shows that when the forces of globalization of economic and financial integration are strong, cross-industry diversification is the superior approach. However, in times of ‘de-risking’ or ‘decoupling’, which could occur if there are geopolitical tensions or global economic crises (or even pandemics), the cross-country strategies would work better.

Myth #5: Index funds offer enough diversification

This depends on the fund or, more precisely, the index it tracks— and many funds do not offer sufficient diversification. This can be determined in a simplified way by looking at what part of the fund (or index) return is generated by what proportion of the constituents. That could be individual stocks, economic sectors, risk factors, countries where companies are based or where they get their revenues from, etc. The smaller the number of stocks/sectors/factors, etc., driving these returns, the higher the level of concentration. For example, below is an extract from Jan-Jun 2023, which shows that the ‘Magnificent 7’ stocks which are the largest in the S&P 500 index in terms of market cap, were responsible for nearly 74% of the index returns over that period. This represents a very significant concentration in large-cap tech companies.

How to deal with all that complexity?

Individual investors can do several things to avoid the traps that come with the above-mentioned caveats. The first step is of course to be aware of them and to have sufficient financial knowledge so that you know what to look for.
– Be mindful of the deceptions of using full-sample correlations. Conditional (tail) correlations would be more useful for allocation decisions — and they should be applied with correspondingly suitable optimization methods.
– If you do not have the possibility or capability, then look for sub-sample correlations corresponding to various economic or monetary regimes or stress vs. non-stress market periods. If possible, use scenario analysis and stress test the correlations to determine how diversified the portfolio would be in case of significant changes.
– Alternatively, tail risks can be hedged (e.g., using derivatives) or mitigated with risk factors that embed short positions or defensive momentum strategies. Consider dynamic risk-based positioning or managed volatility strategies.
– You can track changes in correlations across asset classes and risk factors, as well as their key macro drivers and the sensitivities of the assets you have in your portfolio. This will help you plan a reaction strategy should diversification benefits get reduced.
– Based on your time horizon, goals and constraints, determine which markets to position yourself in and whether cross-industry or cross-country diversification could work best.
– If you invest in funds, make sure their underlying portfolios (or indices) are indeed diversified, not just in name. The number of stocks could mislead you into thinking there is sufficient diversification when in fact it is only 1 or 2 companies or sectors driving the returns.

conclusion

A well-diversified portfolio might seem elusive given that asset correlations are asymmetric and can vary a lot over time and with changes in the economies, financial conditions, and market sentiment. But it is feasible if the individual investor accepts that it would require some work to properly measure and track how various holdings move together in different environments, and how to adjust them over time.

references

1. Page, S., et. al., When Diversification Fails, Financial Analysts Journal, 2018
2. Brixton, A., et. al., A Changing Stock–Bond Correlation: Drivers and Implications, The Journal of Portfolio Management, 2023
3. Viceira, L., et. al., Global Portfolio Diversification for Long-Horizon Investors, Harvard Business School Working Paper, 2018
4. Lombardi, M., et. al., Markets adjust to “higher for longer”, Extract: The correlation of equity and bond returns, BIS, 2023
5. Ilmanen, A., et. al., The Death of Diversification Has Been Greatly Exaggerated, The Journal of Portfolio Management, 2013
6. Mosk, B., et. al., Cross-asset correlations in a more inflationary environment and challenges for diversification strategies, Financial Stability Review November 2022, ECB
7. Attig, N., et. al., Diversification during Hard Times, Financial Analysts Journal, 2023