Decreasing Stock Correlations: Time to Adjust Your Diversification?

If you search the internet for the word “diversification”, you will quickly come across a quote like this:

  • The only free lunch on Wall Street is diversification.

Everybody loves a free lunch.  So, diversification must be the first-stop dining destination for every individual investor, right?  Unfortunately, as I’ve written in the Diversification Series at Mindfully Investing, the prospect of a free meal can lead the undiscriminating investor to a stew of overblown expectations and a portfolio that deflates like a bad soufflé.

The purpose of this post is to check on recent market developments and see what, if any, changes may be needed to the mindful stock diversification approach.  But first, you will need a little background.

Diversification at First Glance

The idea behind diversification is that you should avoid putting all your investment eggs in one basket.  (I just couldn’t resist one more food metaphor.)  If one investment goes rotten, the others still have the potential to meet your investing objectives.  The reasonable sounding assumption behind diversification is that the value of your different investments will not go up and down in lock-step.  The hope is that when some investments are lagging, others are still performing well.  In technical terms, your various investments have to be “uncorrelated” (move randomly to each other) or “negatively correlated” (move in opposite directions) to some degree in order for diversification to provide that free lunch.

Taking a Closer Look

The problem with this logic is that there are virtually no investment asset classes that are consistently negatively correlated.  And investments that are uncorrelated today, may become more correlated tomorrow.  In fact, when the economy goes bad, it’s common for a wide variety of assets to collapse in unison.  This was obvious to anyone who experienced the 2008 financial crisis.  As shown in this graph, typical “flight to safety” havens like long term government bonds (TLT) and gold (GLD) performed relatively well as compared to the S&P500 (^GSPC) stock index.  (Click to enlarge the image.)

But almost all other options performed poorly, including real estate (VNQ), emerging stock markets (VWO), foreign developed stock markets (VEA), and high-yeild corporate bonds (JNK).

Stock Diversification

If you’ve read articles on this website before, you know that a mindful investing approach heavily favors stocks because of their clear superior historical performance, even after considering the risks involved.  Based on current conditions, most long-term young investors are well served by stock-only portfolios.  And portfolios with 80% stocks make sense for most “older” investors, with the remainder in relatively poor-performing but stable cash or bond “ballast”.

Unfortunately, obtaining meaningful diversification with such stock-heavy portfolios is problematic.  As I presented in Article 7.1 and evidenced by the 2008 crash, stock correlations have increased conspicuously in the last 20 to 30 years.  Here’s just one example graph from JP Morgan showing correlation (vertical axis) over time (horizontal axis) for developed markets (DM) and emerging markets (EM):

I conducted an extensive analysis of stock correlations in 2015/2016 and wrote at that time:

  • Diversification across stock types is clearly not a silver bullet that will suddenly reduce the frequency or magnitude of future declines in your stock portfolio with any consistency.
  • At best, you should expect to see a moderate dampening of such declines in some instances as compared to holding an apparently less diversified group of stock funds.
  • Because of the seesaw return performance of various stock types, extensive stock diversification also does not predictably result in consistent and substantial increased returns as compared to holding a smaller less-diversified group of stocks.

Consequently, my overall conclusion was that holding a handful of moderately diversified stock index funds or ETFs is probably all the diversification most people need.  If you add more than that, you are adding expenses and complexity that could just as easily degrade your portfolio’s performance because of the inherent unpredictability of future markets in any given time span.

Times Are A Changin’ ?

I also concluded in that same article:

  • As our investment plans play out, we need to monitor ongoing stock correlations and evaluate whether they’re changing.  Our investment plans should be flexible enough to allow shifts in stock holdings in the event that stock movements become more decoupled in the future.

I’ve come across several news articles recently, for example here and here, indicating that stock correlations are decreasing substantially.  This brings us back to the main purpose of this post: Are these ongoing correlation changes sufficient to alter our mindful stock diversification conclusions?

A Mindful Update

Unfortunately, news articles don’t provide much detail, and I didn’t find any articles that I thought were convincing.  So, I examined recent correlation changes across all the stock types used in my original analysis:

  • U.S. Size (market capitalization)
  • Geography (countries)
  • U.S. Sectors
  • U.S. “Quality” (value vs. growth)

For consistency with that past analysis, I ran several index ETFs for each of these stock types (19 in all) through the Asset Correlation tool in Portfolio Visualizer.  In each case, I selected the longest correlation duration available from the tool (90-day rolling correlations), because long-term investors are interested in long-term correlation trends.  I also selected daily return correlations (instead of monthly or annual), because the tool provides a nice output graph based on daily return correlations.  Here’s a matrix of the most recent 90-day rolling correlations (November 2017) for all possible pairings.  (Click on the image to enlarge.)

You’ll note a few patterns.  Specifically, correlations in the upper left portion of the matrix are still pretty high (generally above 0.5).  These high correlations are all between indices for U.S. value, growth, and size as well as between foreign and domestic markets.  However, the individual U.S. stock sectors in the bottom half of the matrix exhibit some pretty low and even negative correlations with many other indices.  The U.S. sectors with the lowest correlations are consumer staples, REIT (real estate investment trusts), telecommunications, energy, and utilities, which is a notably good outlier.

As I’ve previously pointed out, correlations are dynamic over time, as the JPM graph above illustrates.  This correlation matrix represents only one snapshot for November 2017.  Understanding how correlations change over time is critical to determining any long-term potential diversification benefit.  After examining the time graphs for all these comparisons, I can report that almost all recent correlations are substantially lower than they were just one or two years ago.  Because there are too many correlation graphs to present here, I picked out 12 examples of this pervasive downward trend.  (For a detailed view, hover over the graphic and click on the enlarge  button.)

The downward trends seen in these examples are echoed in almost every correlation chart I generated.  In some cases, the downward trend started as early as 2012, while in others, decreases have only been observed since about 2016.  For variations in value/growth, size, or geography shown in the top two rows of graphs, the correlations are still pretty high.  Accordingly, the correlation declines have been relatively recent and small for these stock types.  However, for the U.S. stock sectors shown in the bottom two rows of graphs, the correlations have trended downward for many years and have declined substantially.  The most extreme case is shown on the bottom right graph, where the utilities and energy sectors went from a near perfect correlation in 2011 to a somewhat negative correlation in 2017.

Time to Revise the Mindful Stock Diversification Plan?

When I last conducted an analysis like this in 2015/2016, I was working with some data sets that were a few years old at that time.  Back then, prior upward trends in correlations appeared to still be in place.  At the same time, many experts were claiming that the rise of passive investing was causing increased correlations in the stock market.  Here’s one example graph (right side) from a Schroeders article:

Put another way, the hypothesis was that increased correlations were a systemic change caused by the rise of passive investing, rather than a cyclical change that was likely to mean revert.  It now seems pretty clear to me that this hypothesis was wrong.  The proportion of passive investing continues to increase and yet stock correlations appear to be uniformly decreasing.  I fell into the same trap as many others.  I assumed that correlation was causation.  The correlation in this case was between the rise of passive investing and the rise in stock correlations themselves.  The more recent data suggest that the ongoing rise of passive investing is not a primary cause of stock correlation changes seen over the last 5 to 6 years.

This is a good reminder of how difficult it can be to be completely rational.  What seems like clear evidence today, can become an irrelevant variable tomorrow.  Consequently, I would like to amend my previous stock diversification conclusions somewhat:

  • All evidence suggests that stock correlations will continue to ebb and flow unpredictably.  Consequently, it would be defensible to add a few additional stock types to the previously recommended “handful” to try to capture the benefits of this unpredictability.
  • Regardless, it’s still true that extensive stock diversification (holding many stock types) does not predictably result in consistent and substantial increased returns or moderation of substantial declines as compared to holding a smaller less diversifed set of stocks.

A Specific New Game Plan

So, what does that all mean in terms of a specific portfolio?  There’s no one specific stock portfolio that will be ideal for everyone.  My own portfolio is very consistent with the entirety of the Mindfully Investing website including on the subject of stock diversification.  So, let’s use my portfolio as a specific example of how a stock investing game plan might be changed because of these recent correlation developments.

My overall portfolio is currently set at the following stock allocations using low-cost index ETFs or similar:

  • 60% U.S. all market index
  • 25% Developed markets (without the U.S) index
  • 15% Emerging markets index

I was never convinced that allocating to U.S. size differences (like a small cap fund) or quality differences (like a value fund) would raise my returns or moderate steep declines much, and these new correlations still support this position.  All the U.S. size and quality indices still have relatively high correlations with other stock types, which is consistent with a long history of relatively high correlations among these stock types.  Yes, correlations of these stock types have decreased some recently, but it’s not enough to make much difference.

In contrast, the large changes seen in the U.S. sector correlations is compelling to me.  The correlations of the consumer staples, real estate, telecommunications, energy, and utilities sectors in particular have swung substantially in the last few years.  Although they may swing back tomorrow, these sectors appear to have a greater potential for periodic divergences from the wider U.S. stock market.

I have some new investable money coming available at the end of the year.  Consequently, I plan to add some new low-cost index ETFs focused on two or three of these stock sectors to further diversify my stock portfolio.  I plan to keep my over all U.S./foreign allocation at 60%/40%.  Accordingly, I will shift some of the new money that was previously ear-marked for U.S. total market index funds to these U.S. sector-focused index funds instead.

How does all this change your stock diversification game plan, if at all?

Leave a Reply

Your email address will not be published.