I was discussing with colleagues recently the market’s change in near term focus from the possibility of a Grexit to the evolution of Federal Reserve monetary policy.
The conversation moved to a discussion of the risks of relying on traditional diversification when we know that correlations are neither stable nor easily predictable.
This reminded me of a recent blog post written by my colleague Bob Collie. The original is repeated below:
Investment would be much simpler if the way that markets reacted to any given event was completely predictable: “if X happens, then Y always follows” would make for an easier life for all of us. But markets are not that well-behaved.
The image above comes from Stas Melnikov and Peter Mortensen of Russell’s risk management group. It plots two different chains of events that can be triggered by a fall in asset prices. On the left is the archetypal reaction: a market decline leading to investor fear and rising demand for US treasuries as a safe haven asset. That represents the classic ‘risk-off’ market move which results in a drop in treasury yields. On the right is an alternative scenario: the same fall in asset prices but no panic-induced flight to safety. In the second case, there is no drop in treasury yields.
There is a whole list of reasons that events might unfold one way or the other. Some of those are quantitative (such as the economic backdrop) while others are more to do with market sentiment. In other words, it depends on the market environment.
This is a prime example of why correlations should not be relied upon as a definitive measure of the interaction between two investments. Correlation is a simple measure of the extent to which two assets tend to move together: a correlation of 1 (or 100%) means that when one goes up, the other always goes up too; if there is no connection at all between price movements then the correlation would be zero. Because it is a simple measure, it is handy as a first-order guide to the relationship. But it is an incomplete description of the way assets interact, especially when relationships are non-linear. Correlation is prone to mislead if used carelessly.
In the example above, the correlation between the price of treasury bonds and other assets is negative during a flight to safety, because Treasury prices rise while other asset prices fall. But during other times, the correlation is more likely to be above zero. Different environments; different interaction; different correlation.
So what does all this mean from the perspective of a scheme trustee?
We know that a negative correlation between bonds and equities (‘flight to panic’) typically amplifies funding level variability. A zero or positive correlation leads to lower funding level volatility for pension funds. Counter-intuitively, the traditional idea of diversification between bonds and equities is not always a good thing for pension funds in the context of their overall balance sheet.
Not only are correlations unstable, they are not easily predictable either. In recognition of this, we avoid an over-reliance on some mathematically-derived strategic asset allocation and prefer a dynamic and adaptive approach to managing balance sheet investment risks.
David Rae, managing director & head of LDI solutions, EMEA, Russell Investments