“Risk is the feeling of disutility caused by uncertainty.” I first heard that definition almost 30 years ago in business school, and it has stuck with me ever since. It’s a novel approach, taking notions of riskiness out of the statistical realm and putting them squarely between our ears. Risk management isn’t really about math. It’s about avoiding an undesirable emotion – fear.
Take the US equity market’s performance last year as Exhibit A: a 22% total return for the S&P 500 with not a single monthly loss… No feelings of disutility there, or fear…
…And this year as Exhibit B: the S&P 500 up 0.3% with drawdowns in two out of three months. Fear has been off the leash for most of 2018 and running around like it owns the place.
Aside from that performance data, the most notable difference between 2017 and 2018 comes down to one word: correlations. Last year, industry sectors and asset classes moved quite independently of each other. This year they are clustered together, limiting the ability of diversification to smooth out the bumps.
We’ve been tracking this data every month since 2009; here’s what we see in the March/April numbers and how they fit into the overall investment landscape:
Point #1: S&P 500 sector correlations are now back to the bad old days.
- At the start of the year, the average 30-day correlation between all sectors of the S&P to the index itself was 0.47. Readers with a long memory will recall this was a normal reading in the 1980s – 1990s.
- For all of 2017, the average S&P sector correlation to the 500 was 0.54 – also within the range of “Old normal”, especially when compared to 2014 (0.77 correlation), 2015 (0.83) and 2016 (0.73).
- Since early February’s market scare, sector correlations have bounced back to the undesirably high levels of the near past. For the 30 days ending April 13th, the average sector correlation was 0.80. That’s a touch lower than March (0.87) and February (0.84), but still very high compared to 2017.
- The CBOE VIX Index tracks sector correlations tightly (0.63 correlation for those two datasets), and we’ve seen this play out with higher overall market volatility since February.
Key takeaway: as much as 2017’s low correlations resembled the bull markets of the 1980s and 1990s, they now look like a temporary anomaly rather than a new regime. US equity markets have returned to their high-correlation ways for a host of reasons (geopolitical, trade/tariffs and cyclical concerns) and we don’t see those abating any time soon.
Point #2: Sectors that are setting – and bucking – the trend.
- There are six sectors that still currently show +0.9 correlations to the S&P 500, even though the average is 0.80. In declining order they are: Technology (0.97), Financials (0.96), Health Care (0.95), Consumer Discretionary (0.95, and remember Amazon is in this group), Industrials (0.94), and Materials (0.91).
- Rate sensitive groups are showing much lower correlations to the market overall: Utilities (0.26), Real Estate (0.51), and Consumer Staples (0.73). Telecomm shows a 0.85 correlation over the last 30 days, the one anomaly in this cluster.
- Energy is the one outlier, with a 30-day correlation to the S&P 500 of 0.76 – lower than average but not tied to the direction of interest rates.
Key Takeaway: there aren’t many places to lower portfolio volatility in a sensible way, but Energy is the best bet here. The fundamental case for the group is reasonable assuming modestly higher commodity prices over the balance of the year. Middle East geopolitics actually gives the sector a tailwind there, unlike any other industry in the S&P 500. Lastly, the group is neither sensitive to interest rates nor a large enough piece of the index (just 6% weighting) to be predestined to decline in a market swoon.
Point #3: Good diversification is really hard to find.
- Non-US equities remained tied at the hip to the S&P 500. The 30-day correlations for EAFE (Europe, Asia, Far East) developed economy equities relative to US stocks is currently 0.92, up from 0.64 at the start of the year. For Emerging Market equities, that number is 0.91, up from 0.45 in January.
- Fixed income correlations to US stocks are a study in extremes. At one end, High Yield corporates show a 0.85 correlation to the S&P over the last 30 days, up from 0.47 at the start of the year. At the other end are long dated Treasuries, where 30 day correlations are -0.58 versus -0.15 at the start of the year. Both observations are currently at their highest relative levels for 2018 to-date.
- If you want to add asset classes with low correlations to a US stock portfolio, the recent data points in two directions. The first is Investment Grade corporate debt (0.03 correlation over the last 30 days). The second is precious metals (gold: -0.18, silver: +0.4). The first is too exposed to rising rates for our comfort. The second option we like a lot more. It’s too early to pound the table on this idea, but rising US budget deficits and inflation are fundamental positives for gold.
Key Takeaway: higher correlations across global asset classes means higher risk for all investors and this limits the amount of new capital that can flow into stocks and bonds. Remember how 2017 set records for US listed exchange traded fund inflows? That was in part due to low correlations. If correlations remain high in 2018, expect to see much lower fund flows into global risk assets than last year.
Our bottom line: “You don’t need a weatherman to know which way the wind blows”, as Bob Dylan sang. The market’s winds have shifted from pleasantly benign (2017) to gusty (now). That has tightened up sector and asset class correlations, putting a chill to the air. That’s fine – investment seasons change, too. Just remember to pack a sweater.