Fear, Volatility, and Correlations

By datatrekresearch in Blog

Human emotions may drive stock market volatility, but sector correlations are the mathematical enablers of both sudden downdrafts and sleepy melt-ups. While sector performance doesn’t always sync up with how well a group tracks the overall equity market, a quick case study from 2007 – 2009 shows the power of correlations:

2007 (a normal year):

• S&P 500 total return: 5.5%
• Best 2 sectors: Energy (+34.4%) and Materials (22.5%)
• Worst 2 sectors: Real Estate (-17.9%) and Financials (-18.6%)

Translation: some groups did quite well, others did not…

2008 (spiking correlations):

• S&P 500 total return: -36.6%
• Best 2 sectors: Consumer Staples (-15.4%) and Health Care (-22.8%)
• Worst 2 sectors: Financials (-55.3%) and Materials (-45.7%)

Translation: everything tanked.

2009 (declining correlations):

• S&P 500 total return: +25.9%
• Best 2 sectors: Tech (+61.7%) and Materials (+48.6%)
• Worst 2 sectors: Utilities (+11.9%) and Telecomm (+8.9%)

Translation: everything worked, but there was a very wide dispersion.

To dig deeper into this topic, today we’ll look at the long run correlations between the S&P 500 and large cap Tech, Financials and Energy. Each has been an area of market interest recently. First, here are the average correlations and the standard deviations of those numbers back to 1999 using 90-day rolling return data:

• Technology: 0.88 with a standard deviation of 0.05 (r-squared of 69% - 87%)
• Financials: 0.86 with a standard deviation of 0.09 (r-squared of 59% - 90%)
• Energy: 0.64 with a standard deviation of 0.23 (r-squared of 17% - 76%)

Takeaway: Tech is structurally the most correlated to the S&P of these 3 sectors over the last 2 decades, Financials aren’t as close as the headline number indicates (almost 2x the standard deviation to Tech) and Energy is simply its own beast.

Now, let’s look at each sector’s correlation time series:

Technology (1999 – Present):

What we see:

• As much as Tech may be a clear leadership group today, its trailing 90-day correlation is 0.92, not far off its long run average of 0.88 and within 1 standard deviation.
• But… Tech sector correlations now (0.92) are actually higher than during the late 1999 dot com melt-up, when they ranged from 0.86 to 0.89.
• Tech’s breakout moment in this cycle was in 2017 (that large drop in correlations towards the right side of the graph), when it meaningfully outpaced the S&P (+39% vs. 22%) for the first time since 2009.
• While the group lost Facebook and Google in September 2018, the drop in correlations in Q4 was more due to the market fearing higher interest rates than any change in index composition.

Bottom line: Tech’s ever-larger weighting in the S&P 500 (24% today) means correlations have likely reached a new plateau, one even higher than during the late 1990s.

Financials (1999 – Present)

What we see:

• Current correlations of 0.87 are essentially the same as the long-run average of 0.86.
• But… just look at all the volatility in correlations from late 2016 – late 2019. That’s the same period as when 3-month/10-year Treasury spreads went from 200 basis points to zero.
• Contributing to (but not entirely explaining) this paradigm shift: losing Real Estate Investment Trusts to their own S&P sector in August 2016. These names were 19% of the Financials sector at the time, not enough of a weighting to wholly remake the group into a pure yield curve play.

Bottom line: Financials can be leadership again (or at least track the broad market more predictably) when bond markets signal renewed confidence in US economic growth.

Energy (1999 – Present)

What we see:

• Current correlations of 0.59 are lower than the long run average of 0.64.
• And… this chart looks nothing like the other two. Energy stocks have not shown reliable +0.8 correlations to the S&P 500 since early 2014, when its sector weight was +10%. It is currently 3.7% of the S&P.

Bottom line: the correlation data combined with the Energy sector’s now-small weighting highlights a less-appreciated reason so many investors have given up trying to time over/under weights in the group. First, general market direction is largely irrelevant, so you have to get the call just right. Second, it’s now not a large enough part of the S&P 500 to make that work/stress worthwhile. Our advice is to always equal weight the sector, purely as a geopolitical hedge.

Final thought: while there are obviously 8 other sectors to consider, the aggregate data from these 3 (41% of the S&P) shows average correlations are either spot on (Tech, Financials) or below (Energy) long run levels. That is a neat and clean explanation for why overall US equity market volatility remains low despite coronavirus fears. You can’t spell “fear” without “correlations” …