US equities are eight years into an equity bull market, which must end. Investors seem to have realised this and are asking how they can protect their equity allocations if and when a correction happens. On a Shiller price-to-earnings basis, US equities have been more expensive on two occasions since 1880 – in 2000 before the dotcom crash, and in 1929 prior to the Great Depression – so such consideration is wise. Potential solutions, however, are in short supply.
To look for answers, the Man Group conducted a study which looked at the seven instances between 1985 and 2016 when the S&P 500 fell by more than 15%, and what might have been the best way to protect an equity portfolio. Contributors included Matthew Sargaison, Dan Taylor and Edward Hoyle. Initially, it examined four passive strategies which should provide crisis protection. These were: long US Treasury bonds, short US credit risk, long gold and long S&P 500 puts. The returns in excess of the one-month Treasury rate – as a proxy for the risk free rate – are shown in figure 1.
All four strategies have strong crisis-period performance on average, but all have drawbacks.
Long Treasury bonds produced an annualised 10.7% excess return over crisis periods but this figure is driven by crises that occurred after 2000. This is a function of stock/bond correlations turning negative at about that time. In the prior century this spread had been predominantly positive. If historical precedent reasserts itself then the utility of long bonds may diminish.
Short credit risk worked well during the global financial crisis where it returned 127%, but experienced variation in other crises. More consistent was the long S&P 500 puts strategy, which delivered double-digit returns in each of the crises, averaging at 41%. The premium expense of rolling the contracts considerably eroded performance, however, which probably makes the strategy a burden for conventional portfolios.
Long gold’s crisis returns were modest (8% on average). The commodity is exposed to idiosyncratic risk unrelated to equity markets, making it an uncertain choice for the future.
In contrast to these passive strategies, our research showed that two dynamic strategies consistently demonstrated their potential for crisis protection. The first is futures time-series momentum. The intuition here is that this resembles a synthetic long-volatility position, something which would benefit in a crisis. We constructed a three month momentum portfolio with exposure to commodities, foreign exchange, equities and fixed income. We then evaluated the effect of different constraints including setting the maximum bound of the net equity position at zero, or limiting the beta of either the equity or the entire portfolio to the S&P 500 to zero. The results can be seen in chart 2, which shows performance across the seven crises corresponding to chart 1.
As expected, restricting the equity exposure is beneficial for performance, but the devil is in the detail. Not allowing the beta to equities to be positive instead of not allowing long equity positions, leads to a worse overall performance without much added benefit during crises based on our simulations (in particular when the cap is applied to the entire portfolio rather than just the equity portion). This suggests that a simple solution may be the most effective, a financial example of the KISS principle: ‘keep it simple, stupid’.
The other strategy is a long-short quality stock portfolio. Our evaluation of quality-minus-junk model returned 43.7% during the crisis periods while maintaining a positive 5.3% across the timeframe.
There is robust intuition behind this. In a crisis, investors experience a flight-to-quality impulse. Within equities, this can manifest itself in a move towards profitable companies, with records of growth and underleveraged balance sheets, and investor friendly. In a similar manner to trend following, where we found that different capping mechanisms can have different crisis alpha properties, our work found that portfolio construction and changing the metric used to measure quality affects performance.
Quality stocks have had a near zero correlation with time-series momentum strategies over time. This may mean that, not only are these strategies effective hedges, but they could represent appropriate complements to each other. A futures trend approach needs time to react to market patterns and therefore may be of benefit in prolonged crises. A quality stocks strategy, on the other hand, has more of a ‘sit and wait’ character, performing better in general and providing crisis relief when turbulence strikes.
This means that, when the two are combined, the proportion required to make a material impact on a portfolio is maybe less than one might assume. Our study found that a proportion of just 10% in this combination (leaving 90% in long equities) would reduce the annualised loss during crises by eight percentage points.
Otto van Hemert is head of macro research at Man AHL