Birds of a Feather Fall Together

By in
Birds of a Feather Fall Together

If February’s sell off gave you flash backs to 2008 or 1987, there’s a good reason. Asset class and sector correlations spiked this month, just as they do in every violent market decline. It is the most trying aspect of a sudden price drop: there is simply no place to hide within equities. You either sell or you ride it out.

Exacerbating that feeling of a plummet into the abyss: until late January, sector correlations had been dropping for over a year. This trend actually started right after the 2016 elections, and 2017 saw the lowest industry sector correlations to the S&P 500 of the last decade. That was an underappreciated factor in the low volatility melt up last year; lower correlations among the sector components of the S&P 500 naturally reduced day-to-day swings for the index as a whole. This made for a very smooth ride higher.

To see how sharp a swing correlations took, consider the following (correlation data from

  • The average price correlation for the 11 sectors of the S&P 500 over the last month was 83.7%. For the 3 months prior, they were 37 – 49%.
  • The correlation between EAFE (non-US Developed Economy) equities and US stocks was 95.0% over the last month. The range for the 3 months prior: 34-72%.
  • Emerging markets got sucked into the selloff as well: price correlations to the S&P 500 last month were 87.9%, up from 46-59% the three months before.
  • While high-grade corporate bonds didn’t really track stock prices over the last month, high yield bonds did. Their price correlation to the S&P 500 rose to 72.1% from 35-57% the three months prior.

How closely did the selloff in long term Treasuries line up with the decline in stock prices? The math from the last month says “Not much”, but we take little comfort in that. The one-month correlations are -14.9%, but we think that understates the relationship since the decline in equities only occurred in the last 2 weeks.

Now, this may seem like dull inside-baseball stuff, but correlations will be very important in 2018. Two reasons why:

  • We probably aren’t going to see S&P sector correlations return to their 2017 lows. We’ve outlined in prior notes our fundamental view that the story for US equities has changed from checkers (lower rates, higher earnings) to 3D chess (higher rates, deficit spending, dollar weakness). That is the recipe for higher volatility – last week was the appetizer course – and therefore more risk-off downturns. The upshot is that correlations will also likely whipsaw in 2018, feeding the downside moves as they converge.
  • Correlations are a critical driver of risk parity portfolios – the lower asset price correlations are, the more leverage they tend to take. Recall that risk parity is the term of art for bond-stock portfolios that add leverage to juice returns from what is a very plain-vanilla investment style. Now, they won’t make dramatic changes to their portfolios if correlations return to 2017 norms – these are battleships and supertankers, not speedboats. But a lasting move higher for correlations may push them to delever (sell assets) or reconfigure their approach (sell bonds to buy stocks or vice versa).

The key takeaway: if correlations change, some very large investors may have to change their market exposure. Last year was about as good as it gets: bonds and stocks both up, low correlations and volatility allow for leverage to maximize returns. This year will likely be quite different.

One final thought to complete the discussion: what role did exchange traded funds, such as those that mirror the S&P 500, play in the sudden spike in correlations? The SPY ETF saw $23 billion in outflows last week, essentially equal to the total redemptions from all US equity funds ($25 billion). Yes, that selling likely helped push correlations higher. But recall that 2017 was a banner year for ETF fund flows and correlations fell dramatically.

The most accurate observation we can make with that conflicted history is that ETFs may contribute to short-term correlation clustering, but the longer term record says there is less of a relationship. Still, it is one more thing to consider if you buy into our view that 2018 will have more equity price volatility.