We all know that putting uncorrelated stocks, or uncorrelated assets into a client’s portfolio will reduce risk and limit drawdowns. But what does “uncorrelated” really mean, and how much does it help? Better to know now than be surprised in the future.
Think about two men, or women, each flipping a coin. The results are certainly uncorrelated. Statistics show that there is a 25% chance one will show heads and the other tails, and another 25% that the opposite person will show tails and heads. That leaves 25% when both show heads and 25% when both show tails. If we think about it in terms of stock prices, prices will move the same direction 50% of the time, even if those stocks have nothing whatsoever in common.
When we construct a stock portfolio, it is traditional that we look for those companies that offer diversification, that is, stocks that don’t move the same way at the same time. We typically measure that using correlations.But of course, they do have something in common – they are driven by the economic health of the country. At a minimum, we should expect any two stocks, picked arbitrarily, to move in the same direction more than 50%. Let’s look at some examples.
Business as Usual
We’ll start by assuming that more diversification is better, and that the simplest way to find the relationship between any two stocks, or a stock and the broad index, is to find their correlation. This is a simple from for Excel, but remember that correlations use the daily returns, not the prices.
The correlation, which ranges from +1 to -1, tells us how similar the movement of one stock is to another. A value of +1 means it’s identical, -1 that it’s exactly the reverse. Those cases don’t happen. The value 0 says that there is no relationship between the price movement. My own assessment is
+0.50 to +1.00 Very strong correlation
+0.20 to +0.50 Positive correlation
-0.20 to +0.20 No relationship
-0.20 to -0.50 Clear inverse relationship
-0.50 to -1.00 Unlikely for happen
We see strong positive correlations between interest rate vehicles, such as a 2-year and 5-year Treasury note. We also see it when there is a crisis, and investors all redeem their funds at the same time. It’s the money that moves the market.
As an example, let’s look at four different stocks, General Electric (GE), Bank of America (BAC), Merck (MRK), and Amazon (AMZN), each quite different. Table 1 shows the correlations against each other and against SPY (the SPDR ETF) for the 20 years beginning 1998.
Table 1. Correlations from 1998.
The averages on the right show that the strongest correlation is against SPY, and less when one stock is compared with another. We should expect that because the SPY is an average and picks up the general direction of the market. These four stocks are included in that average.
We don’t usually think of price moves in terms of correlation, instead we think about whether stocks move up and down together. If we just look at the days stocks moved the same way, we get a much higher relationship. In Table 2, we see how stocks moved compared to GE. I think it better represents the way we think about stock movement.
Looking at the average correlation over a long period hides a lot of information. How much did the correlation change during that time? How did it react during a crisis, such as we saw in 2008? Can we pick stocks to diversify a portfolio from the long-term correlations?
During July and August 2017, the market moved slowly higher. We can call it business as usual. The dominant news was increasing tensions between the U.S. and North Korea, and Amazon buying Whole Foods and then lowering prices. SPY prices are shown at the top of Chart 1 and the rolling 20-day correlations are in the bottom panel. The rolling correlations give us a much better understanding of how price patterns change.
When the SPY rises, we see the correlations increasing. This happens at the beginning of July and the end of August. During the sideways periods, the correlations move towards zero. Each stock moves according to its own fundamentals. These periods alternate, giving us the sense that we have good diversification.
Correlations During a Crisis
It’s not the ordinary market that is the problem, but the extremes. If we’re good at diversifying a portfolio, we have been protected during the financial crisis of 2008. But we weren’t. Chart 2 shows the rolling 20-day correlations of the four stocks against SPY for 2008. Notice that Merck and Amazon both had periods of low correlation, but the overwhelming picture is that correlations hovered around 0.80, showing extremely high similar movement.
Chart 2. Correlations during the 2008 financial crisis.
Price Movement in a Crisis
It may be easier to see if these occasional, low correlations meant that a stock was rallying while others were declining. In Chart 3 we can see the prices of the four stocks and SPY. Despite the drop in the correlations of Merck, the price chart shows that it declined more than the other stocks during 2008, and all four stocks decline more than the index, SPY. Then choosing stocks based on either long-term or short-term correlations did not prove to be helpful during a crisis. Table 2 shows how the average correlations varied during the three periods we’ve looked at.
Chart 3. Prices of the four stocks and SPY, adjusted to 100 on January 1, 2008.
Table 2. Rolling correlations, against SPY, during three representative periods.
How Do You Select Stocks to Protect Risk?
We’ll be forced to use common sense to create a portfolio that protects an equity investment during a crisis. But first we need to be specific about the crisis, and that is a problem. We know that the attack of 9/11 could have been protected by defense stocks and bonds (the flight to quality). The recent North Korean risk also calls for defense stocks. Then there was the financial crisis of 2008. Again, risk-free bonds would have helped.
Do we need to consider anything else? An attack on our power grid? An energy shortage? An assassination? It’s going to be necessary to decide which events would cause a large move in the stock market, what stocks or other assets would protect us, and what portion of the portfolio should they represent.
The 1987 Stock Market Crash
While not the same as 1929, the stock market crash in October 1987 was the largest since the Great Depression. The Dow dropped 27.2% on the Monday and Tuesday of October 18th and 19th. At the same time, bond futures rallied (Chart 4) and gold flopped around (Chart 5).
Chart 4 Bonds rally while stocks fall.
Chart 5 Gold reacts as an immediate hedge, but then collapses.
It’s clear that the flight to quality means government interest rates, but gold failed after years of being touted as an important hedge.
The Internet Bubble
The next event was the internet bubble, which saw the stock market go into a 3-year decline after an incredible rally in Nasdaq stocks in the later 1990s. Chart 6 shows the S&P futures with bond futures, although the S&P did not receive the brunt of the losses. Again, we see bonds as a good hedge.
Chart 6. The Internet bubble bursts. Investors run to bonds.
Then there was 9/11. We were all there, one way or another. Stocks plunged, bonds soared. Only the defense stocks, Lockheed-Martin (LMT), Raytheon (RTN), and Northrop-Grumman (NOC), rallied. In Chart 4, we’ve added Boeing (BA) to show that aerospace is not the same as defense. At the same time, Chart 8 shows that bonds did not offset the equity losses. The initial reaction was that yields rose, then ended the year higher (futures lower).
Chart 7. Defense stocks rally during the 9/11 terrorist attacks.
Chart 8. US bonds did not provide a hedge during the 9/11 period.
The 2008 Financial Crisis
Although a bit late and a bit sloppy, bond futures finally rallied in response to the severe sell-off in late 2008. Before that, it provided a reasonable hedge as the S&P steadily declined (see Chart 9).
Chart 9. Bond futures provided a reasonable hedge during the financial crisis of 2008, although its reaction at the worse time was sloppy.
Although we’ve seen more conflicts during the past two decades, Iraq and Afghanistan seem less important in light of the tension and escalation of rhetoric between the U.S. and North Korea. The defense stocks seemed to have anticipated the crisis well before it became news. Chart 10 shows the movement of the three defense stocks and Chart 11 shows how bonds reacted.
Chart 10. The defense stocks react to N Korean tensions.
Chart 11. Bonds mostly track the S&P during N Korean missile tests.
The combination of the S&P and bonds has served well as a conservative portfolio for years. During a financial crisis or swings in the U.S. dollar, bonds have been the safe haven. During periods of strong economic activity, stocks give the best returns. During a crisis, stocks drop and bonds rise, just as we like.
But times have changed. Terrorism and geopolitical threats have become too common. Neither bonds or gold show any consistency in protecting portfolio risk. However, defense stocks have filled that gap.
To show that defense stocks are not just a stop-gap, Chart 12 shows the history of those stocks, along with SPY, adjusted to zero at the beginning of the chart. It’s clear that defense stocks are just as likely to perform well, even while having their unique hedging quality.
Chart 12. History of defense stocks from 1998, compared to SPY.
Then, a combination of a broad market index, or diversified individual stocks, bond futures, and defense stocks will be the better choice. But in what proportion?
By the Numbers
Again, we’ll start by looking at a few cases. When the stock exchange reopened on 9/17/2001, the SPY closed down 4.21 points, while LMT, RTN, and NOC closed up and average of 6.03 points. We can’t do this in percent because prices of those stocks have been adjusted for splits. Based on the price moves, a portfolio of 58% SPY and 42% defense would fully hedge the crisis.
For the 2008 financial crisis we need to look at futures. Again, back-adjusting makes it impossible to use percentage returns. From July 2007 through December 2008, the S&P dropped from about 1400 to under 600, 800 points, although we know it was equivalent to about 50%. One contract at $50 per point, give a total loss of $40,000 per contract. Yoiks! At the same time, bonds rallied (more or less) from 55 to 85, settling at 77, up 22 points, equivalent to a gain of $22,000. Then a portfolio that had two contracts of bonds for each one of the S&P would have been fully hedged.
If we put together the S&P, bonds, and defense stocks, we get:
27% S&P futures + 54% bond futures + 19% defense stocks = 100% investment
Where the S&P and bonds are futures. Bonds can vary based on maturity and whether they are cash, futures, or ETFs. Using only S&P and bond futures, from 1987, in the proportion 27-54, we get the results in Chart 13. Note that there are no declines in 1987, and after the internet bubble. There is no decline for 9/11 and a sideways period that ends in 2003. There is volatility in 2008 but no significant drawdown. Net gains are 25% better than a portfolio of 60% stocks and 40% bonds, but more important, the risk is far lower.
Chart 13. Comparing a portfolio of 33% stocks and 66% bonds with the traditional 60% stocks and 40% bonds.
Adding 19% defense stocks to the portfolio of stocks and bonds presents an intellectual dilemma (see Chart 14). Performance is not as good as the portfolio without the defense stocks. The returns are slightly lower and the risk is slightly higher. The question becomes “Do I believe that defense stocks with be important in the future?” My vote is “yes.” If you don’t like 19%, then any added allocation to defense stocks will reduce exposure to some future price shocks. Good management is about anticipation.
Chart 14. Portfolio of 27% stocks, 54% bonds, and 19% defense stocks.
We could even go a little farther and include cyber-security stocks, if we could find one or two that reacted correctly to the major hacking events. The value would only be seen if the equity index, the S&P, dropped after yet another announcement of a security breach.
The other challenge is to reverse the weighting of the industry’s 60% stocks and 40% bonds to a more conservative 33% stocks and 66% bonds. While we tend to stay with the traditional recommendations until it’s too late, reducing leverage will capture more of the gains from the bull market of the past six years. Isn’t it time to change the paradigm?
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.