“Understanding the recent rise in correlations” by William J. Coaker Jr. was originally published in Investment Management Consultants Association’s Investments & Wealth Monitor. To access more articles, visit IMCA’s Web site: www.IMCA.org.
Investors who increased allocations to international stocks, emerging markets, real estate, hedge funds, high-yield bonds, and natural resources during the previous decade did so at least partly because these investments’ correlations to U.S. stocks, and to each other, had been low in the past. Unfortunately, the correlations increased significantly in recent years. As a result, an expected reduction in risk did not occur, and in the 2008 bear market investors suffered much larger losses than expected. This article addresses the following:
• Which correlations have increased substantially in recent years; which correlations have remained low
• Why the correlations between many assets have risen sharply over the past five years
• What factors may cause correlations to change in the future and recommendations on how investors can utilize factors that impact correlation to improve asset allocation decisions
Table 1 is a correlation matrix for 20 asset classes covering five distinct time periods:
1. 1970 or the inception date of an asset class through 1999
2. 2000 to 2004, which shows how correlations were changing in the first half of that decade
3. January 2005 to October 2007, which shows asset class correlations before the current financial crisis
4. November 2007 to December 2009, which shows their correlations during the current financial crisis
5. Full five-year period from 2005 through 2009
To download Table 1, click here.
Which Correlations Have Risen
Some correlations have risen sharply in recent years, and some have remained relatively low.
S&P 500 to international stocks, emerging markets, high-yield bonds, and real estate. Table 1 shows that the correlations of the S&P 500 to international stocks, emerging markets, high-yield bonds, and real estate have increased significantly in the past five years, from roughly 0.5 to 0.8 and even 0.9. The correlations of most of these assets to the index had risen significantly before the crisis, and they moved notably higher during the crisis.
The index’s relationship to natural resources was low before the crisis but moved up when trouble appeared, from –0.10 to 0.55.
The index maintained a low correlation to U.S. bonds, global bonds, U.S. Treasuries, Treasury inflation-protected securities (TIPS), and gold, both before and during the crisis. Its correlation to hedge funds remained relatively constant—around 0.7—both before and during the crisis.
Other U.S. equity styles to international stocks, emerging markets, real estate, and high-yield bonds. During January 2005–October 2007 other U.S. equity styles—large value, small growth, etc.—also experienced significant increases in correlations to international stocks, emerging markets, real estate, and high-yield bonds. These correlations then moved higher during the crisis.
Other U.S. equity styles had low correlations to natural resources before the crisis but moved up during the crisis, from around zero to 0.5.
Other U.S. equity styles maintained low correlations to U.S. bonds, global bonds, U.S. Treasuries, TIPS, and gold both before and during the crisis. Other equity styles’ correlation to hedge funds has remained relatively high, in the 0.7 to 0.8 range.
International stocks to U.S. stocks, emerging markets, hedge funds, high-yield bonds, and real estate. The correlations of international stocks to the S&P 500, emerging markets, and hedge funds moved notably higher during 2000–2004, then shot up into the 0.9 range during the crisis.
International stocks’ correlations to high-yield bonds and real estate remained low before the crisis but moved up significantly, from around 0.30 to 0.82 and 0.79, respectively, when trouble emerged. International stocks’ correlation to natural resources remained around zero before the crisis but moved up to 0.59 during the crisis. International stocks have maintained a low correlation to U.S. bonds, global bonds, U.S. Treasuries, and TIPS, both before and during the crisis.
Emerging markets to U.S. stocks, international stocks, high-yield bonds, and natural resources. The correlation of emerging markets to international stocks rose from 0.48 during 1986–1999 to 0.81 during 2000–2004; then moved in lock step at 0.95 during the crisis. Emerging markets’ correlation to the S&P 500 jumped from 0.51 before 2000 to 0.77 during 2000–2004, then leaped to 0.87 during the crisis. Emerging markets’ correlation to high-yield bonds also moved higher, from 0.35 before 2000 to 0.65 during 2000–2004, then to 0.84 during the crisis. Emerging markets’ correlation to natural resources moved up from 0.05 before 2000 to 0.36 during 2000–2004, then jumped to 0.64 during the crisis. Emerging markets always have had some extra correlation to hedge funds, ranging from 0.73 to 0.88 depending on the time series. But emerging markets’ correlations to U.S. bonds, global bonds, U.S. Treasuries, TIPS, and gold has remained relatively low, even through the crisis.
Real estate to U.S. stocks, international stocks, emerging markets, and high-yield bonds. Real estate’s correlations to the S&P 500, other U.S. equity styles, international stocks, emerging markets, and high-yield bonds historically had been moderate or even low but rose significantly during the crisis. For example, real estate’s correlation to international stocks had been 0.36 before November 2007 and rose to 0.79 during the crisis. Real estate maintained low correlations to U.S. bonds, global bonds, U.S. Treasuries, natural resources, TIPS, and gold before and through the crisis.
High-yield bonds to U.S. stocks, international stocks, emerging markets, hedge funds, and real estate. The correlations of high-yield bonds to the S&P 500, other U.S. equity styles, international stocks, emerging markets, hedge funds, and real estate traditionally had been moderate to low before the crisis, and all these correlations rose during the recent period of crisis. High-yield bonds’ correlation to U.S. bonds, global bonds, and U.S. Treasuries remained low before and throughout the crisis. But high-yield bonds’ correlation to hedge funds, natural resources, and TIPS all rose appreciably during the crisis.
Natural resources to emerging markets, hedge funds, and, to a lesser extent, U.S. stocks, international stocks, and high-yield bonds. Natural resources had a very low, almost-zero correlation to every type of asset before 2005, and these correlations remained quite low until 2005. During January 2005–October 2007, natural resources’ correlations to emerging markets, hedge funds, and gold moved up, but the correlations with all other assets remained low. During the recent crisis, natural resources’ correlations to emerging markets and hedge funds rose further, to around 0.7. Natural resources’ correlations to the S&P 500, international stocks, and high-yield bonds also went from near zero before 2005 to around 0.5 during the crisis. Natural resources had low correlations to U.S. bonds, global bonds, U.S. Treasuries, real estate, TIPS, and gold before the crisis, and these low correlations held during the recent crisis.
Hedge funds to emerging markets, international stocks, high-yield bonds, natural resources, and U.S. stocks. Hedge fund assets surged throughout the 2000s, partly due to hedge funds’ low correlations to most assets. Unfortunately, as hedge funds became popular, their correlations to several assets began rising, particularly for international stocks during 2000–2004 and for natural resources and gold during January 2005–October 2007. During the crisis, hedge funds had correlations of 0.87 to emerging markets, 0.82 to international stocks, 0.80 to high-yield bonds, 0.75 to natural resources, and 0.73 to the S&P 500. Hedge funds have had consistently low correlations to U.S. bonds, global bonds, and U.S. Treasuries, but their correlations to TIPS and gold has edged up in recent years.
U.S. bonds, global bonds, U.S. Treasuries, TIPS, and gold have maintained low correlations to other assets and to each other. The correlations of many assets rose sharply during 2005–2009, but U.S. bonds, global bonds, U.S. Treasuries, and, to a somewhat lesser extent, TIPS and gold, have maintained their historically low correlations to all other assets, with a few exceptions. For example, the correlation between U.S. and global bonds was 0.35 during 1985–1999 and was still just 0.48 during January 2000–October 2007, but it jumped to 0.79 during the crisis. The correlation for gold and TIPS rose from around 0.20 to 0.55 during the crisis, and the correlation for TIPS and global bonds has been rising for the past 10 years.
Why Correlations Have Risen
Numerous factors explain why the correlations between many assets have risen.
The financial crisis. Table 1 shows that the correlations between many assets jumped significantly during the crisis. The correlations of the S&P 500 to international stocks, emerging markets, and real estate; international stocks to emerging markets and high-yield bonds; and emerging markets to high-yield bonds and natural resources, and several others, all shot up during the crisis. During the crisis, severe macro worries overwhelmed micro and fundamental factors, leading to a rise in the correlation of many assets as nervous investors sought to sell a wide array of risky positions, reduce their risk and leverage profiles, and increase their liquidity.
But the correlations of many assets had been rising significantly before the crisis. Specifically, the correlations of U.S. stocks to international and emerging market stocks; international to emerging market stocks as well as hedge funds; and emerging market stocks to natural resources all had risen substantially before the onset of the crisis. Hence, other factors in addition to the crisis appear to be instrumental in explaining why the interaction between certain asset classes has tightened.
Free trade. A total of 253 companies in the S&P 500 show foreign sales as a percentage of all their revenues in their quarterly reports. For those 253 companies, foreign sales accounted for 47.9 percent of company revenues as of June 2009, up from 43.6 percent just three years earlier (Standard and Poors 2009). These companies tend to be larger, multinational companies, so if data were available for all 500 companies in the index, foreign sales would be somewhat lower, perhaps around 40 percent. Table 2 shows the percentage of foreign revenues for 28 large companies domiciled in the United States. The average foreign sales for those 28 companies is 71 percent (Standard and Poors 2009).
Growth in gross domestic product (GDP) in the United States averaged 3.3 percent during 1950–1999, but slowed to 2.6 percent during 2000–2009. Europe also experienced sluggish growth in the 2000s, and Japan has languished with very slow growth for two decades. In other words, the vast majority of the developed world experienced relatively slow economic growth in the 2000s. Comparatively, emerging markets grew more than 8 percent annualized from mid-2003 through mid-2008 (JPMorgan 2010). As growth slowed in the developed world and surged in the emerging world, companies in developed countries significantly increased their sales to emerging markets.
At the same time, China and many other emerging markets built their strategies for fast economic growth on exports to the developed world, primarily to the United States. The effect of increasingly free trade was that the correlations between U.S., international, and emerging markets stocks rose significantly.
Hedge funds. In the 2000s hedge fund assets exploded, rising from $189 billion in 1999 to $826 billion by 2003, then surging to $2.14 trillion by 2007 (but falling to $1.46 trillion in the 2008 bear market; by June 30, 2010, hedge fund assets totaled $1.64 trillion) (see www.barclayhedge.com). The combination of the explosion of assets that hedge funds had to invest, plus their discretion to invest in different types of investments in any part of the world, and the high turnover and momentum orientation of many of their strategies, led to a crowding effect that caused significant jumps in correlations for many of the asset classes in which hedge funds invest.
Figure 1 shows that the growth in hedge fund assets coincided with the time period in which the correlations rose substantially, as shown in table 1. As hedge funds grew, it became more difficult for them to execute their investment strategies without significantly impacting the prices of the assets in which they invested. In 2008 hedge funds declined 19.0 percent, according to the Hedge Fund Research Index (www.hedgefundresearch.com), shocking investors. Before 2008, hedge funds posted positive returns every year since 1990.
High-frequency trading. Computerized, high-frequency trading (HFT) also exploded during the past decade, and HFT often operates on a mix of momentum factors. Kemp (2010) argued that the growth of momentum trades further raised the volatility and correlations of a wide array of risky assets. Smith (2010) concluded that HFT is having a large impact on the microstructure of equity trading dynamics and is resulting in a compression of correlations.
Exchange-traded funds. As shown in figure 2, ETFs totaled just $37 billion in 1999. ETFs grew to $227 billion by 2004 and surged to $780 billion by 2009 (Grantham, Mayo, and Otterloo, personal communication 2010; Automated Quantitative Research [AQR], personal communication 2010). Barclays Capital (2010) recently stated that “since 2005, equity correlation has had a close relationship with the increased ETF volumes relative to the volumes in the underlying stocks.” ETFs give investors a faster way to increase or decrease exposure to an asset class than other forms of trading. As use of ETFs has grown in recent years, asset prices have been marked by more sudden risk-on, risk-off moves. The growth of ETFs has added to the momentum and volatility of markets and helped raise the correlations between assets.
Derivatives. The use of derivatives, such as futures and swaps, has exploded over the past decade. Derivatives now total more than $500 trillion in notional value, which is more than 10 times global annual production (Davies 2010). Just as with ETFs, derivatives give investors the ability to increase or decrease exposure to an asset class very quickly. The growth of derivatives, as well as HFT, ETFs, and hedge funds, has been instrumental in the rise of asset correlations as well as the increased volatility in the financial markets.
The Future of Correlations and Recommendations for Investors
High correlations are a sign that investors are worried the crisis has not passed. In periods of trouble, macro worries tend to dominate the micro. There are exceptions, such as the Greek debt crisis earlier this year, but during a crisis—and particularly during a crisis as severe as the current one—systematic risks overwhelm idiosyncratic risks.
Currently, the macro concerns are numerous, and significant: a slowing U.S. economy, housing remains worrisome, high rates of delinquent borrowers, high unemployment, enormous government deficits and debts, greater regulation, uncertainty and worry among both business and consumers, extraordinary government intervention into the economy, deflationary pressures and inflationary threats, and prospects for higher taxes.
Longer-term worries include the future of the eurozone, trade and currency wars, and the potential for a bond crisis that triggers a collapse of the fiat money system and ushers in a new world order.
We’re still trying to ward off depression and financial tragedy. Elevated correlations may persist until the crisis has passed. Until the crisis shows more signs of being behind us, investors should watch their allocation to assets with elevated correlations, particularly to assets with high betas and high volatility.
Differences in fundamentals, where they exist, eventually will cause correlations to decline. While the financial crisis has been instrumental in the rise of asset class correlations, and one day the crisis will have passed, most of the other factors that caused correlations to rise are systemic and not temporary. Investors appreciate the benefits of ETFs and derivatives. HFT is here to stay. Investors still are embracing hedge funds. Companies domiciled in developed countries are increasing their sales in emerging markets.
High correlations may persist until the current crisis has passed, and remain higher than they were before the crisis, but differences in fundamentals between asset classes eventually will become too compelling for investors to ignore. The price of an asset can become separated from its fundamental value for many years, but fundamentals ultimately determine price. Hence, if correlations between assets remain high amid differences in fundamentals, the correlations will, eventually, decline.
Due to the significantly greater use of momentum strategies, when correlations do begin to decline they may do so quickly and sharply. Hence, investors need to remain disciplined to their asset allocation strategies so that they reap the benefits of diversification if correlations decline.
Further, in the case of another extreme selloff, investors should consider altering their asset allocations to buy good assets on the cheap. In 2008, emerging markets and high-yield bonds were the worst-performing asset classes. In 2009, they were the best, and they have outperformed in the first eight months of 2010.
Stress-testing and scenario-based asset allocation. According to modern portfolio theory (MPT) and the theory of normally distributed returns, the October 2008 decline was a three-standard-deviation event that should occur roughly once every 9,000 months (i.e., 750 years). Even more astonishing was the blowout in yields in October 2008. According to MPT, the yield spread between investment-grade bonds and U.S. Treasuries that month should occur once in the lifetime of the universe. Correlations tend to increase significantly during major declines, which obviously occur more frequently than MPT would predict, and investors should stress-test asset allocation decisions to take this fact into account.
Investors also should consider designing scenario-based asset allocations because of current deflationary pressures and inflationary threats. Deflationary pressures are present in the form of consumers who are reducing debt and increasing savings, and inflationary threats loom from governments and central banks printing trillions of dollars.
If deflation takes hold, high-quality bonds would do well but stocks would not, and their correlations would remain very low.
But if inflation takes hold, the correlation between stocks and bonds could remain low or it could rise. On the one hand, stocks offer inflation protection, which bonds do not. Hence, with inflation bonds would do poorly but stocks could hold up reasonably well, in which case the correlation would remain low.
But with inflation stock returns tend to be disappointing because investor sentiment is low, causing their price-earnings ratios to decline.
In short, if inflation takes hold when price-earnings ratios are high, the returns for both stocks and bonds probably would disappoint, and the correlations would be higher than normal. But if inflation takes hold when price-earnings ratios are low, bonds still would do poorly, stocks would have a better chance to earn decent returns, and correlations would remain relatively low.
Be aware of common factor exposures among different asset classes. Coaker (2006, 2007) found that real estate, U.S. bonds, high-yield bonds, and hedge funds have been more closely correlated to value investing than growth, likely because of their common exposure to debt levels, the quality of debt, the availability of credit, the willingness to borrow, and interest rates.
Before late 2007, debt levels were rising substantially, the quality of debt was declining, access to credit was easy, and interest rates were low. Hence, from 2000 through mid-2007 asset classes with extra sensitivity to and dependence on credit conditions—value style investing, real estate, bonds, high-yield bonds, and hedge funds—all earned pretty good returns. But when credit conditions reversed in late 2007, the result was the most severe economic contraction since the Great Depression, and most of these asset classes with extra sensitivity to credit incurred larger losses than other risky assets. Their correlations also rose much higher than their historical averages.
In short, investors need to comprehensively understand the factor risks that are common among seemingly unrelated asset classes. In this case, value stocks, real estate, high-yield bonds, and hedge funds all share extra sensitivity to credit conditions.
Factors that May Cause Correlations to Change (Again)
Impact of free trade and fiscal and monetary policy on equity correlations. The correlation of the S&P 500 to international stocks has risen sharply as free trade has expanded. This correlation probably will remain considerably higher than it was before 1990. Two factors could cause correlation to decline: 1) protectionist measures and 2) individual government’s fiscal and monetary policies, as well as their approaches to solving deficits and debts. If their approaches are significantly different, the correlation between U.S. and international stocks could decline. The correlation between U.S. large and small stocks is high, ranging from 0.80 to 0.96. If protectionist sentiment rises and trade barriers gain traction, this correlation could fall because multinational companies obtain nearly 40 percent of their revenues from sales overseas.
Impact of emerging markets embarking on domestic-driven growth strategies on equity correlations. The correlation of emerging markets to the S&P 500 has soared lately. That’s because emerging markets focused on an export-driven growth strategy and U.S. companies increased sales to emerging markets. However, now that the U.S. consumer is heavily in debt, emerging markets are focusing more on domestic-driven growth as well as trading with each other. Further, the fundamentals of emerging markets are good—low debt, high savings, and high economic growth—so they are in a good position to execute such a strategy successfully. If they are successful, the correlation between emerging markets and the S&P 500 could fall. That said, the correlation may not fall substantially because U.S. multinationals derive a significant amount of revenue from emerging markets. Further, the correlations of emerging market countries to each other, which already have jumped over the past five years, would remain elevated as their trade with each other continues to expand.
Impact of government debt on the correlation between stocks and bonds. The correlation of bonds to every other asset historically has been quite low, and it remained low throughout the current crisis. But the correlation between bonds and stocks could rise under certain conditions. For example, government debt has increased sharply, and if a crisis developed related to government’s ability or willingness to meet debt obligations, bonds would do poorly, and stocks could get hit hard as well.
Factors that impact TIPS returns.TIPS have had a very low correlation to stocks, but they may not provide the inflation protection investors expect. The factors that affect the returns for TIPS are beyond the scope of this article. Kim (2008) provides a complete discussion on the factors that determine the returns of TIPS.
Impact of growth in emerging markets on the demand for natural resources. Natural resources and emerging markets used to have no correlation to each other, but they do now. Several emerging market countries, such as Brazil and Russia, are resource-rich, while others, such as China and India, are resource-poor but rich in labor. As companies in developed markets moved more of the cost of their operations to China and India, the wealth in those countries rose. That led to increased demand for natural resources, boosting Brazil, Russia, and other resource-rich countries. As a result, the correlation between emerging markets and natural resources rose from 0.00 before 2000 to 0.57 during 2005–2009.
Impact of demand for natural resources on the correlations to other assets. Natural resources used to have very low to negative correlations to all other assets, but that has changed lately. Natural resources’ correlations to U.S. stocks, international stocks, emerging markets, high-yield bonds, and hedge funds were all near 0.00 through 1999, then climbed to 0.40 to 0.65, depending on the asset, during 2005–2009. High rates of economic growth and enormous populations in China and India are causing a surge in demand for natural resources. Hence, based on demand the correlation of natural resources to other assets may continue to be higher than before 2005. But the correlations may fall if the other side of the equation—supply of natural resources—becomes impaired and a scarcity of goods emerges.
Impact of government policy on the risk, returns, and correlations on investments. To avoid a second Great Depression, governments worldwide asserted themselves aggressively into vast segments of the private sector, including the banking, finance, housing, insurance, and auto industries. In the future, government involvement in the private sector is likely to be much greater, and taxes appear poised to rise, perhaps significantly. With government asserting much greater control in the private sector, investors should consider the implications that government actions—including regulation, mandates, taxes, and protectionist measures—are likely to have on future returns, risks, and the correlations between asset classes.
William J. Coaker, Jr., CFA®, CFP®, is the senior managing director of equity investments at the University of California. He earned a BS in accounting from Loyola Marymount University and an MBA from Golden Gate University. Contact him at email@example.com.
1 The author calculated correlations using monthly returns during 1970 or the inception of an asset class index through December 2009. The following indexes were used for each asset class (the date in parenthesis is 1970 or the inception of the asset class index):
S&P 500: Standard and Poors 500 Index (1970) Large Growth: Russell 1000 Growth Index (1972); Fama and French (1970–71) Large Value: Russell 1000 Value Index (1972); Fama and French (1970–71) Mid-Growth: Russell Mid-Growth Index (1986) Mid-Blend: Russell Midcap Index (1979) Mid-Value: Russell Mid-Value Index (1986) Small Growth: Russell 2000 Growth Index (1979)
Small Blend: Russell 2000 Index (1979); Ibbotson (1970–78)
Small Value: Russell 2000 Value Index (1979)
International Stocks: MSCIEAFE Index (1970)
Emerging Markets: MSCI Emerging Markets Index (1986)
Real Estate: National Association of Real Estate Investment Trusts (NAREIT) Index (1972)
Bonds: Barclays US Government/Credit Index (1970)
High Yield Bonds: Barclays High Yield Corporate Index (1983)
Global Bonds: CG World Global Bond Index (1985)
Treasury Inflation-Protected Securities: Barclays TIPS Index (1997)
Natural Resources: Goldman Sachs Commodity Index (GSCI) (1970)
Hedge Funds: Hedge Fund Research Fund Weighted Index (1990)
Gold: Goldman Sachs Gold Total Return Index (1970)
Barclays Capital. 2010. Barclays Index Volatility Weekly. Barclays Capital Equity Research (July 26), available for subscribers at http://www.scribd.com/doc/35061853/Barclays-Index-Volatility-Weekly-20100726.
Coaker, William J. Jr. 2006. The Volatility of Correlation: Important Implications for the Asset Allocation Decision. Journal of Financial Planning 19, no. 2 (February): 58–69.
Coaker, William J. Jr. 2007. Emphasizing Low Correlated Assets: The Volatility of Correlation, Journal of Financial Planning 18, no. 9 (September): 52–70.
Davies, Roy. 2010. Gambling on Derivatives: Hedging Risk or Courting Disaster? (May 11), available at http://projects.exeter.ac.uk/RDavies/arian/scandals/derivatives.html.
JPMorgan. 2010. Guide to the Markets (3Q2010): 37, available at http://www.jpmorganfunds.com/cm/Satellite?UserFriendly URL=insidemarkets&pagename=jpmfVanityWrapper.
Kim, Jane J. 2008. Treasurys Tied to Inflation Gain Appeal; Yields on So-Called TIPS Have Risen in Market Tumult; Watching the Break-Even Rate. Wall Street Journal (October 8): D1.
Kemp, John. 2010. Rising Correlation and Computer-Driven Trading, Reuters.com (July 13), available at http://blogs.reuters.com/great-debate/2010/07/13/rising-correlation-and-computer-driven-trading/.
Smith, Reginald. 2010. Is high-frequency trading inducing changes in market microstructure and dynamics? Quantitative Finance Papers from www.arXiv.org, available at http://econpapers.rep Kim, Jane J. 2008. Treasurys Tied to Inflation Gain Appeal; Yields on So-Called TIPS Have Risen in Market Tumult; Watching the Break-Even Rate. Wall Street Journal (October 8): D1. ec.org/paper/arxpapers/1006.5490.htm.
Standard and Poors. 2009. Foreign Sales by U.S. Companies Continue to Rise, Standard and Poors.com. Press release (July 14, 2009), available at http://www2.standardandpoors.com/spf/pdf/index/20090714_SP500-FGN-SALES-2008_PR.pdf.