Why the correlation bubble isn’t going to burst

By Felix Salmon
August 19, 2011

Everybody knows that correlations go to 1 during a crisis. But markets and crises move at many speeds. Rodrigo Campos quotes one technical analyst today as saying that during the latest bout of market volatility, “We were moving over a three and a half day period like we were during the flash crash, just more orderly. It was totally irrational.”

Paul Amery today takes a bigger step back, looking at stock-market correlations over the past five years or so. The chart is instructive, I think:


The methodology here is a bit complicated — you start with the implied volatility of stock index futures and then strip out the implied volatility of single-stock options. And towards the end of each contract’s life things can go a bit haywire — especially if the contract is expiring in January 2009, right in the middle of the crisis. But the big picture seems clear: correlations are not only very high, but they’re on a steady uptrend, too. What we’re seeing right now is not some kind of unusual spike: it’s entirely in line with how the stock market has been trading for years now.

JP Morgan, in a research report last year, looked even further back, using a different methodology.


The big thesis of the report was that correlations had overshot, and were likely to come back down; that didn’t happen. Instead, the trend has only continued since then: correlations have been increasing for a good 20 years at this point, and are now at levels exceeding those seen even in the aftermath of the 1987 crash, when portfolio insurance products forced high correlations on the market as a whole.

There are many reasons why correlations are high and rising. The rise of high-frequency trading is one; the rise of ETFs is another. Both of those are here to stay.

And then there’s the broader economy. Here’s how JP Morgan puts it:

A significant driver of correlation between stocks is the prevailing macroeconomic environment. During periods of high macro uncertainty, stocks prices are largely driven by macro factors such as economic growth, unemployment, interest rate changes, inflation expectations, etc. Therefore, during changes in macroeconomic regimes, stock prices tend to move in unison leading to a high level of correlation.

On top of that, macroeconomic uncertainty tends to lead to stock-market volatility, and stock market volatility in turn is associated with higher correlations. So the more uncertain the macro environment, the higher correlations are likely to go. JP Morgan might be right that we’re in a “correlation bubble”, but if we are, I don’t think it’s going to burst any time soon, just because macro uncertainty is going to remain very high for the foreseeable future.

What are the implications of this? For one thing, expect your broad-based S&P 500 index fund to give you much less diversification benefit than it would have done ten years ago; similarly, you can expect it to be much more volatile than it would have been ten years ago, too. Essentially, indices are behaving more and more like risky individual stocks, with the proviso that they can’t go to zero.

On top of that, it’s going to remain very hard to make money as a stock picker in this market. It’s easier than ever to play the indices, thanks to ETFs, which now account for 40% of all exchange-traded activity in equities. And if you want extra risk, that’s easy, too: just play bank stocks instead of the broad index. Because of their inherent leverage, they always fall more than the index when it’s going down, and rise more when it’s going up. But if you’re carefully doing the Graham-and-Dodd thing, poring over balance sheets and trying to find value, make sure you have a long time horizon and aren’t liable to get run over by the correlation steamroller. Your carefully-picked stock will, to a first approximation, do exactly the same thing as all the other stocks are doing, and if you’re planning on waiting for corporate fundamentals to assert themselves, be prepared to wait a long time.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

It may be worthwhile here to mention that market-cap weighted indexes are not really unbiased. I read recently that Apple accounts for something like 40% of the value of the S&P 500.

An investor might do better by focusing on, say, the bottom 100 of the 500.

Posted by daniellc_boca | Report as abusive

The Graham-and-Dodd thing has probably moved into relative-value long-short plays. That’s one area where correlation likely helps you (with caveats, of course).

Posted by absinthe | Report as abusive

“I read recently that Apple accounts for something like 40% of the value of the S&P 500.”

Nowhere close. It may have accounted for 40% of the *growth* in the S&P500 over a given period of time, but that kind of statistic is suspect because the total gains for the gainers total much more than 100% — they are simply offset by a similarly large number of losers.

Can somebody please define how “correlation” is measured in this context? Over what time period? It seems to me that “correlation” is a lot like “volatility”. If you measure your time frame in months or years, the correlation breaks down. Why should I care if the individual ticks are highly correlated or not?

Posted by TFF | Report as abusive

I think Daniellc was thinking that Apple had grown to roughly 20% of the NASDAQ 100 before it was repegged earlier this year…

Posted by CDN_Rebel | Report as abusive

The Graham-and-Dodd thing has always been a long time horizon deal. If it made profits quickly, then everybody would do it.

As far as correlation goes, I wouldn’t blame ETF’s. ETF’s merely take the place of Index funds and I don’t think there was a huge new adoption of indexing in 2008, when correlation spiked. Rather, in 2008, tons of investors who did have their money sitting in indexes through thick and thin put their money instead in Treasuries. The sensitivity of passive investors to macro news has created the long-term, 2-year correlation. It is also the reason 3-month correlation went down in 2009 and has gone back up again reasonably. It has to do with index investors going into and out of the market, not with ETF’s or HFT.

If we reach “normal” times again, where passive investors don’t suddenly bail en masse into indexes (or go back en masse into indexes), correlation will come back down. Until then, the companies that are priced too cheaply will earn interest on their extra free cash flow and eventually the earnings and cash will overwhelm their price. It may take awhile for that to happen, but it wouldn’t take a decade or anything like that.

Posted by mwwaters | Report as abusive

Uh, reasonably means recently. Short-term Correlation has gone back up because macro news has turned very sour again. I really wish these comments had an edit function.

Posted by mwwaters | Report as abusive

Golden Networking has created an instructional DVD for executives and professionals in high frequency trading, entitled, “The Speed Traders Workshop 2012”. This 4-disc DVD set walks professionals and non-professionals through the main issues, challenges and opportunities practitioners face to consistently capture alpha using fast computers. The DVD uses non-mathematical terminology and provides insights to successfully implementing speed trading while avoiding minefields. It is the ideal reference for professionals in the world of alternative investments. Visit http://bit.ly/znIl3E for more information

Posted by kaylabi | Report as abusive