History since the turn of the millennium has been marked by several themes: an ever growing dependency on smart phones, the recent trend towards populism and the economic rise of China. Negative stock-bond correlations rarely make the list.
Admittedly not quite rising to the significance of the smart phone, but this is a big deal from the narrow perspective of the asset allocator. Since 2000 stocks and bonds have tended to move in opposite directions. This propensity towards negative correlation has made bonds a reliable hedge against equity risk. Whether this trend continues is key for how investors build portfolios.
It is important to recognize that stock-bond correlations have not, as a matter of course, always been negative. In fact, over the long term stock-bond correlations average roughly zero. That said, the average masks two very distinct periods. See the chart below.
From the 1980s through the bursting of the tech bubble, correlations were reliably positive, averaging 0.50. This was a period when traders anxiously awaited every weekly money supply print and had to divine Federal Reserve (Fed) policy without explicit communications.
The second period began in early 2000. In addition to the end of the tech bubble, that year also marked the start of the slow growth regime that we are in today. Since 2000 correlations have often been negative.
Why the shift?
While the lack of independent economic cycles argues against too strong of a conclusion, stock-bond correlations have tended to co-move with inflation and monetary conditions. During the past 25 years, there has been a tendency for correlations to be higher when Fed policy is tighter. With the Fed tightening monetary conditions for the first time since the crisis, stock-bond correlations may be heading higher.
This leaves the question of why investors should care. Does it really make a difference if stock-bond correlations are -0.4, -0.2, 0 or 0.2? As it turns out, it matters quite a bit when building a multi-asset portfolio.
Correlation decides allocation
Looking at a simple asset allocation, a theoretical allocation to long-dated U.S. bonds (+20 years) fluctuates from as low as 3% to as high as 25% based on changes to the risk model, i.e. correlation of different asset classes. Even with no change in return estimates, a significant change in correlation can induce massive shifts in allocations.
When stock-bond correlations are presumed to be negative, portfolio construction favours traditional Treasury bonds—particularly long-dated ones—as a good source of both carry and diversification. When stock-bond correlations are positive, other hedges—notably cash—may be preferable.
This is an under-appreciated dynamic. In an environment where expected bond returns are already low, the ability to hedge equity risk is the key driver of a bond’s weight in a portfolio. Most investors are good at quoting yield; it may be time to become just as familiar with this less obvious but equally important number.