Clik here to view.
One of the most vexing problems in investment management is that diversification can disappear when investors need it most. In fact, correlations—the degree to which returns on different assets tend to move in the same direction at the same time—tend to increase in down markets, especially during crashes. Studies have shown this effect to be pervasive across a large variety of financial assets and sectors, including individual stocks, equity country markets, global equity industries, hedge funds, currencies and international bond markets.1 Interestingly, while the increase in correlations was particularly acute during the fourth quarter of 2008 and the first quarter of 2009, most of these studies were published before the crisis. Yet, the failure of diversification during the crisis seemed to surprise investors.
Not only do correlations increase on the downside, they also tend to significantly decrease on the upside. This asymmetry is the opposite of what investors want. Indeed, who wants diversification on the upside? During good times, shouldn’t we seek to reduce the return drag from diversifiers?
Many investors still don’t fully appreciate the impact of correlation asymmetries on portfolio efficiency, (that is, the degree to which a portfolio targets the best possible tradeoff between expected risk and expected return) and in particular, on their exposure to loss. During sell-offs, diversified portfolios may have greater exposure to loss than more concentrated portfolios. One 2009 study showed that during the 2008 financial crisis, a portfolio diversified across U.S. stocks, U.S. bonds, international stocks, emerging markets stocks and real estate investment trusts saw its equity beta rise from 0.65 to 0.95 and unexpectedly underperformed a simple 60% U.S. stocks, 40% U.S. bonds portfolio by -9 percentage points.2
We took an in-depth look at what drives the stock-credit, stock-bond, stock-hedge fund and stock-private asset correlations and the implications for multi-asset investing.3
The magnitude and inescapability of the failure of diversification across markets may surprise investors. Our goal is to encourage practitioners to take action on this issue.
International Diversification
The correlation between U.S. equities and international equities illustrates our approach. We examined how correlations changed between January 1970 and June 2017. Results were sorted based on 1-month U.S. market returns, ranging from the worst months for U.S. stocks (1st percentile) to the strongest rallies (99th percentile). In a normal case, we should expect perfect symmetry between upside and downside correlations. Also, correlations should gradually decrease as we move toward the extremes, or “tails,” of the distribution of results.
What we found instead was that actual correlations across different market environments differed substantially from the normal case. When U.S. stocks were rallying (the 99th percentile of all monthly results), the correlation with non-U.S. stocks dropped to -17%. However, during the worst monthly sell-offs in U.S. stocks (the 1st percentile), the correlation with non-U.S. stocks rose to +87%. This asymmetry reveals that international diversification worked, but only on the upside. See “Non-U.S. vs. U.S. Stocks,” this page.
Image may be NSFW.
Clik here to view.
Behavior of Risky Investments
We found similar results across relatively risky asset classes when we compared correlations during rallies and sell-offs. See “Key Risky Investments,” p. 68. Again, diversification failed in the worst 1% of monthly periods, across styles, sizes, geographies and alternative assets. Essentially, all of the return-seeking building blocks that asset allocators typically use for portfolio construction were affected.
Image may be NSFW.
Clik here to view.
Shifts in the underlying environment, often referred to as “regime shifts,” may explain this widespread risk-on/risk-off effect. But, what causes these shifts? A partial answer is that there may be structural changes in the macroeconomic fundamentals—such as economic growth, monetary policy, inflation and/or labor market conditions. Also, we surmise that investor sentiment plays an important role. Financial markets tend to fluctuate between a low volatility state and a panic-driven, high volatility state.4
In normal markets, different fundamentals are expected to drive differences across returns on risky assets. During panics, however, investors often “sell risk,” irrespective of differences in fundamentals. Fear appears to be more contagious than optimism. It’s part of human nature to react more strongly to bad news than to good news.
True Source of Diversification?
When market sentiment suddenly turns negative and fear grips markets, government bonds almost always rally because of a flight-to-safety effect.5 In a sense, duration risk—a measure of the sensitivity of bond prices to changes in interest rates—may be the only true source of diversification in multi-asset portfolios. Therefore, the expected stock-bond correlation is one of the most important inputs to the asset allocation decision.
Unlike results for correlations across risky assets, our study found that the profile of stock-bond correlations across different market conditions is a highly desirable one. Bonds tended to decouple with stocks in bad times and become positively correlated with stocks in good times. See “U.S. Stocks vs. U.S. Treasuries,” p. 69.
Image may be NSFW.
Clik here to view.
However, the stock-bond correlation is difficult to estimate and can change drastically with macroeconomic conditions.6 Earlier studies have shown that when inflation and interest rates drive market volatility more than business cycle and risk appetite considerations, the stock-bond correlation often turns positive.7
Over the last decade, central bank stimulus and declining rates have artificially pushed valuations higher in both stocks and bonds, in our view. This type of “sugar high” can unwind quickly when
policy normalizes unexpectedly. The “taper tantrum” in 2013, after the Federal Reserve suggested it would begin reducing its extraordinary liquidity support for the markets, provided a good example, as it affected stocks and bonds negatively at the same time. Starting valuations also matter and can compound the effect of policy changes. The higher the valuations in both stocks and bonds, the more fragile their correlation tends to be.
To illustrate how bond sell-offs can lead to a positive stock-bond correlation, we reversed the conditional correlations. We estimated the stock-bond correlation as a function of U.S. bond market performance instead of U.S. equity performance. See “U.S. Treasuries vs. U.S. Stocks,” p. 69. We found that the correlation profile wasn’t as desirable as when we conditioned on stock returns. Although correlations were generally low, when bonds sold off, stocks often sold off at the same time.
Image may be NSFW.
Clik here to view.
Ultimately, investors should remember that stocks and bonds both represent discounted cash flows. Unexpected changes to the discount rate—the assumed interest rate used to value future cash flows—or in inflation expectations can push the stock-bond correlation into positive territory, especially when other conditions remain constant.
A Broader Look at Hedge Funds
Beyond traditional asset classes, investors have increasingly looked to alternative investment vehicles, such as hedge funds, for diversification. However, our comparison of seven hedge fund styles versus U.S. stocks showed that all of the styles exhibited significantly higher correlations during stock market sell-offs compared to equity rallies, including so-called “market neutral” hedge funds. See “Hedge Fund Styles,” p. 70.
Image may be NSFW.
Clik here to view.
A simple explanation could be that most hedge fund strategies are short volatility. Some are also short liquidity risk, which is akin to selling an option.8 Another study observed that “a wide range of hedge fund strategies exhibit returns similar to those from writing a put option on the equity index.”9 A related study used a regime-switching model to measure hedge fund correlations and market betas over time. It found that the average jump in correlations among hedge fund strategies across financial crises was +33%.10
What About Private Assets?
Over the last few years, institutional investors have significantly increased their allocation to privately held assets. Money has flowed into these asset classes in part due to their perceived diversification benefits. However, the returns reported for private assets suffer from a smoothing bias. In fact, studies have found that their diversification advantages are almost entirely illusory.11 On a mark-to-market basis, these asset classes historically have been exposed to many of the same risk factors that drove stock and bond returns. Not only was the true equity risk exposure of private assets higher than implied by their reported returns on average, but also their exposures to sell-offs were much higher. See “Private Assets Quarterly vs. Rolling Annual Data,” p. 70.
Image may be NSFW.
Clik here to view.
After the smoothing bias was removed, private assets had exposure to credit risk, which explains why, in part, their rolling annual correlations to stocks was quite high during equity sell-offs. Credit risk didn’t truly diversify equity risk in times of market stress. And, even more than for hedge funds, liquidity risk contributed to the asymmetry.
Opportunities for Asset Allocation
We believe that investors should avoid the use of full sample correlations when constructing or adjusting their portfolios. Or at least, they should stress-test their performance assumptions. Scenario analysis, either historical or forward-looking, should take a bigger role in asset allocation. There’s a wide range of portfolio optimization methodologies that directly address non-normal risks during sell-offs.
These analytics are widely available, but they’re too often used on a “post-trade” basis, after portfolio construction has taken place. Instead, we believe investors should put tail-aware tools at the center of their “pre-trade” decision-making process. Doing so may reveal that equity regions, styles and sectors, as well as credit instruments and most alternative investment vehicles, don’t diversify against broad equity risk as much as it might seem on the surface. We’re not arguing against diversification across traditional asset classes, but investors should be aware that traditional measures of diversification may belie exposure to loss in times of stress. In our view, investors should calibrate their risk tolerances (against their return opportunities) accordingly.
We also believe that significant emphasis should be put on the stock-bond correlation and whether it will continue to be negative going forward. Significant shocks to interest rates or inflation can turn this correlation positive. Under such conditions, strategies that use leverage to increase the contribution of bonds to overall portfolio risk may experience unexpected losses.
Lastly, we believe investors should look beyond diversification to manage portfolio risk. Tail risk hedging (with equity put options or proxies), risk factors that embed short positions (a technique used in anticipation of the decrease in value of a security) or defensive momentum strategies (used to de-risk before losses accumulate) and dynamic, risk-based strategies that potentially reduce exposure to loss all may provide protection in a more robust way than traditional diversification.
The good news is that tail risk-aware analytics, as well as hedging and dynamic strategies, are now widely available to help investors seek to manage environments in which traditional portfolio diversification techniques are ineffective.
—A version of this article was originally published in Financial Analysts Journal (2018).
—This material does not constitute a distribution, offer, invitation, recommendation, or solicitation to sell or buy any securities; nor is it intended to serve as the primary basis for any investment decision.
Investors should seek independent legal and financial advice, including advice as to tax consequences, before making any investment decision. Past performance is not a reliable indicator of future performance. All investments involve risk.
MSCI shall have no liability whatsoever with respect to any MSCI data contained herein. This report is not approved, reviewed or produced by MSCI.
Endnotes
1. See, for example, Andrew Ang, Joseph Chen and Yuhang Xing, Downside Correlation and Expected Stock Returns (2002); Andrew Ang and Joseph Chen, Asymmetric correlations of equity portfolios (2002); Yongmiao Hong, Jun Tu and Guofu Zhou, “Asymmetries in Stock Returns: Statistical Tests and Economic Evaluation,” The Review of Financial Studies, Vol. 20, Iss. 5 (September 2007); François Longin and Bruno Solnik, “Extreme Correlation of International Equity Markets,” The Journal of Finance, Vol. 56, No. 2 (April 2001); Miguel A. Ferreira and Paulo M. Gama, Correlation Dynamics of Global Industry Portfolios (2004); Anne-Sophie van Royen, Hidden Risks of Hedge Funds (2002); Vikas Agarwal and Narayan Y. Naik, “Risks and Portfolio Decisions Involving Hedge Funds,” The Review of Financial Studies, Vol. 17, No. 1 (Spring 2004); Philipp Carl Hartmann, Stefan Straetmans and Casper de Vries, “Heavy tails and currency crises” (2010); and Lorenzo Cappiello, Robert F. Engle and Kevin Sheppard, “Asymmetric Dynamics in the Correlations of Global Equity and Bond Returns,” Journal of Financial Econometrics, Vol. 4, Iss. 4 (2006).
2. Martin Leibowitz and Anthony Bova, “Diversification Performance and Stress-Betas,” Journal of Portfolio Management, Vol. 35, No. 3 (Spring 2009), at pp. 34-40.
3. For more details on our study methodology, see Sébastien Page and Robert Panariello, “When Diversification Fails,” Financial Analysts Journal, Vol. 74, No. 3 (Third Quarter 2018), at pp. 19-32, www.cfainstitute.org/en/research/financial-analysts-journal/2018/faj-v74-n3-3. Refer to Appendix B at www.cfapubs.org/doi/suppl/10.2469/faj.v74.n3.3 for data sources and full correlation profiles.
4. See, for example, Chang-Jin Kim, “Unobserved-Component Time Series Models With Markov-Switching Heteroscedasticity: Changes in Regime and the Link Between Inflation Rates and Inflation Uncertainty,” Journal of Business & Economic Statistics, Vol. 11, Issue 3, at pp. 341-349 (1993); Manmohan S. Kumar and Tatsuyoshi Okimoto, “Dynamics of Persistence in International Inflation Rates,” Journal of Money, Credit and Banking, Vol. 39, No. 6 (September 2007); James D. Hamilton, “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle,” Econometrica, Vol. 57, No. 2 (March 1989), Thomas Goodwin, “Business-Cycle Analysis With a Markov-Switching Model,” Journal of Business & Economic Statistics, Vol. 11, Iss. 3 (1993), Rob Luginbuhl and Aart de Vos, “Bayesian Analysis of an Unobserved-Component Time Series Model of GDP with Markov-Switching and Time-Varying Growths,” Journal of Business & Economic Statistics, Vol. 17, Iss. 4 (1999).
5. See Les Gulko, “Decoupling,” Journal of Portfolio Management, Vol. 28, No. 3 (2002), at pp. 59-66.
6. See Craig B. Wainscott, “The Stock-Bond Correlation and Its Implications for Asset Allocation,” Financial Analysts Journal, Vol. 46, Iss. 4 (July/August 1990); Lingfeng Li, “Macroeconomic Factors and the Correlation of Stock and Bond Returns,” Yale ICF Working Paper No. 02-46; AFA 2004 San Diego Meetings (2002); Gulko, ibid.; Sami Vahamaa, Magnus Andersson and Elizaveta Krylova, “Why Does the Correlation Between Stock and Bond Returns Vary Over Time?,” Applied Financial Economics (2008); Lieven Baele, Geert Bekaert and Koen Inghelbrecht, “The Determinants of Stock and Bond Return Comovements,” Review of Financial Studies, Vol. 23, Iss. 6 (June 2010); and Nicholas Johnson, Vasant Naik, Sébastien Page, Neils Pedersen and Steve Sapra, “The Stock-Bond Correlation,” Journal of Investment Strategies, Vol. 4, No. 1 (2014).
7. Johnson, ibid., at pp. 3-18.
8. See Vineer Bhansali, Bond Portfolio Investing and Risk Management, McGraw-Hill Education (2010).
9. See Agarwal and Naik, supra note 1, at pp. 63-98.
10. Monica Billio, Mila Getmansky Sherman and Loriana Pelizzon, “Crises and Hedge Fund Risk,” Yale ICF Working Paper No. 17-14 (Revised 2012).
11. See Niels Pedersen, Sébastien Page and Fei He, “Asset Allocation: Risk Models for Alternative Investments,” Financial Analysts Journal, Vol. 70, No. 3 (May/June 2014), at pp. 34–45.