Return correlations for the major asset classes have been ticking higher in 2018, based on the median of pairwise relationships for rolling one-year periods via a set of exchange-traded products. Although the median correlation is still well below levels reached last year, the upward trend this year is conspicuous, suggesting that diversification benefits generally may soften in the months (years?) ahead.
The median correlation for the major asset classes ticked up to 0.362 through yesterday’s close (July 11), based on daily returns for the latest one-year rolling period (252 trading days). That’s just slightly below the year-to-date high of 0.363 in May. The median correlation has been moderately higher in recent years, reaching the 0.45-0.50 range. Based on the trend this year, it’s reasonable to assume that the median correlation will continue to rise. (Correlation data ranges from -1.0 (perfect negative correlation) to zero (no correlation) to +1.0 (perfect positive correlation.))
Focusing on how correlations have evolved from the perspective of the US stock market shows that a key part of this year’s general increase in correlations has been driven by a tighter link between stock markets across the world. Notably, correlations between Vanguard FTSE Developed Markets (VEA) and Vanguard FTSE Emerging Markets (VWO) have posted higher correlations recently vs. Vanguard Total US Stock Market (VTI), as shown in the chart below. By contrast, the correlation between the US stock market (VTI) and US fixed-income securities via Vanguard Total Bond Market (BND) has remained mostly steady and middling relative to recent history (based on daily returns).
For a deeper review of correlations, let’s turn to the complete dataset for all the major asset classes, based on the following proxies:
Let’s begin with a look at all the pairwise correlations, based on daily returns for the current one-year trailing window. As the table below indicates, most of the correlations are positive. A few pairs, however, post negative correlations – the deepest negative relationship is between US stocks (VTI) and US bonds (TIP and BND).
For another perspective, the next table shows how correlations stack up for daily returns over the trailing five-year period. Once again, the deepest negative relationship is between US stocks (VTI) and US bonds (BND).
Finally, let’s consider one other perspective: this time using one-year returns (252 trading days) to calculate correlations for the trailing five-year period through yesterday. By this definition, the deepest negative correlations are now between foreign stocks in developed markets (VEA) and US bonds (BND and TIP), as shown in the next table below. One takeaway here is that it’s crucial to estimate correlations using a time window that’s appropriate for your portfolio strategy and investment horizon.
If there’s one broad theme in the numbers it’s that there’s a relatively wide range of correlations below 1.0 across the major asset classes. In turn, that’s a sign that robust opportunities are available, albeit in varying degrees, for enhancing risk and return by designing diversified portfolios – an opportunity that’s widely labeled as the only free lunch in investing.
Historical correlations, of course, are only a guide for designing and managing asset allocation. The next step is developing intuition about expected correlations. The obvious starting point for that task, however, is reviewing the historical record.
A New Book From The Capital Spectator:
Quantitative Investment Portfolio Analytics In R:
An Introduction To R For Modeling Portfolio Risk and Return
By James Picerno