- The correlation between asset classes lies at the very core of strategic asset allocation and the search for improved risk-adjusted returns in multi asset portfolios
- Our analysis reveals that the level of inflation and the volatility of inflation have been the most influential drivers historically to the equity/bond relationship in the US
- A new regime? As output gaps close and inflation edges upwards we expect a higher rolling 5yr equity/bond correlation than investors have been used to for most of the past decade
- Rising term premium in US Treasuries may also reflect investors’ belief that the diversification benefits and safe haven utility of bonds are reducing
- We expect the rolling five year correlation to edge further towards zero but given the on-going structural forces weighing on inflation we do not see markets returning to the strong positive correlation regime of the 1970s to late 1990s
Rolling 5-year correlation US Large Cap Stocks–10y nominal US Treasuries (1932–2017)
Why does the correlation matter?
- Any significant shift in the equity/bond correlation ‘regime’, has potentially significant implications for the behavior of multi asset portfolios, for multi asset return expectations and for the portfolio optimization process.
- Shifting correlations also impact potential returns. In theory, the correlation between asset classes should be reflected in embedded risk premia and in expected returns. This simply reflects that rational investors expect higher returns during periods when correlations are high to compensate for lower diversification benefits, and accept lower expected returns when correlations are low because the diversification benefits are more significant.
The bottom line: investment implications
What proportion of a portfolio to allocate to specific assets is the essential quandary facing all multi asset investors—a quandary further complicated by the deliberate role central bank Quantitative Easing programs have played in supporting all risk assets and distorting the correlations that play such an important role in the asset allocation process.
On our analysis, the long run correlation of large cap US equities to 10yr nominal US treasuries is 0.08. We therefore believe that long term investors should probably start with the presumption of zero correlation of stock and bonds.
But recognizing that correlations are dynamic and time varying is essential to efficient portfolio construction. This variability is captured in our own asset allocation modelling using correlation matrices. We do not attempt to call daily changes to investor risk aversion or the equity-bond correlation, but instead focus on the potential for regime change in the statistical relationship between equities and bonds over multi-year periods based on clear evidence historically.
Looking forward, we see inflation ticking higher as global output gaps close and as wage growth rises from its current very low base in the developed world. This is exactly what should be happening at this point in the cycle. We see this modest repricing of inflation as a support not a threat to corporate profitability and to equity prices. We view the probability of a violent shift higher in global bond yields as unlikely in the context of powerful demographic drivers and the on-going expansion of central bank balance sheets globally.
Against this backdrop we expect the rolling 5yr US equity/bond correlation to also edge higher. By definition, this is likely to mean that bonds are not as effective a portfolio diversifier as they are when the equity/bond correlation is strongly negative. But importantly we do not currently expect a return to the sort of strongly positive correlation regime that would likely necessitate major asset allocation rebalancing for investors without a more meaningful and sustained pick-up in inflationary pressures and macroeconomic volatility.