Rising correlation and stock volatility
Mon, Nov 19, 2012 | Jared Woodard
In 2011 and early 2012, one of the big themes was the elevated correlation among individual stocks. In the “risk on, risk off” environment, investor appetites for stocks were indifferent to the fundamentals of particular companies: trades were all about the macroeconomic risks.
You can get a sense for the major swings in stock correlation by looking at the three month correlation among the Dow Jones Industrial Average components since 2008 – see the bottom panel in fig. 1. One surprising result is that the extent to which stocks traded together during the debt ceiling panic and the worst phase of the European banking crisis in 2011 was actually greater than during the 2008 market decline.
Fig. 1. 1-month and 3-month Dow Jones Industrial Average realized correlation.
Source: Yahoo!, Condor Options
The one month estimate in the top panel is noisier, but it shows changes in stock co-movement more quickly. One month Dow correlation fell to 19% this spring and again to 23% in October, but has spiked higher more recently to nearly 40%. To make sense of what different correlation levels mean for stocks, look at the attached scatter plot of daily Dow Jones price changes at different levels of trailing one month correlation.
Fig. 2. 1-month DJIA realized correlation and daily stock returns.
Source: Yahoo!, Condor Options
The idea to take from this chart is that volatility and correlation are positively related. It is intuitive that when investors are trading assets based not on the fundamentals of specific companies but based on some other market-wide factor, index returns will make larger daily swings.
Investors can also take some comfort that, since 2008, stock returns at correlation levels around 0.40 have not been especially worrisome – mean daily returns are still clustered around 0%, and the tails are not extreme. Traders who are deciding whether to increase the size of portfolio hedges should keep an eye on correlation, though – if markets rally in the short term but realized correlation increases, we will be more likely to see higher volatility ahead.
Tags: correlation, dia, djia, Volatility



November 19th, 2012 at 8:50 am
So, in some sense, we are experiencing rising covariance. Is high covariance bad? (Traditional CAPM would say that it is since diversification isn’t as helpful, right?)
November 19th, 2012 at 10:31 am
Yes, correlation just is standardized covariance. It’s bad for CAPM portfolios since assets intended to diversify stop doing so.
November 19th, 2012 at 1:42 pm
Great charts–particularly Fig 1!…you reference Yahoo for the data…can you share how you created that data series? Or is that proprietary…
Thanks
Glen
November 19th, 2012 at 2:13 pm
Hi Glen, I used the quantmod and PerformanceAnalytics packages and some sloppy code in R.
November 20th, 2012 at 2:47 pm
Great post. I particularly like the second chart which shows that high correlation doesn’t necessarily correlate with negative returns, but just an increase in the volatility of the returns.
November 24th, 2012 at 2:09 pm
Where can I get the correlation charts that you display above.
November 24th, 2012 at 2:13 pm
The charts were created using data from Yahoo and code in R.