The words of 2016 (part 4) : Correlations

This is the final post in a series of four on the key words which characterize 2016 from an economic and market perspective. Previous posts discussed ‘surprises’, ‘disconnect’ and ‘regime change’. This post covers ‘correlations’.

Investors have a love/hate relationship with correlations. On the one hand, historically well-established correlations between market behaviour and economic factors, make investment decision making easier. On the other hand, if everything becomes highly correlated, the benefit from portfolio diversification goes down. What’s more, considering that diversification comes at a price (implementation costs and particularly the cost of researching an ever broader set of asset classes), an increase in average correlations between asset classes entails hard dollar costs and not only an opportunity costs in terms of reduced diversification benefits.

2016 has seen some great examples of correlation plays (the charts which follow come from our Review of 2016). The markets for US Treasuries and German Bunds were highly correlated during the bulk of the year. The black line shows the change in the spread on a year to date basis. The fact that this cumulative difference was so small for so long shows that virtually any move in the US yields was reflected in a similar move in German yields and vice versa. This also meant that the correlation of this spread with the EUR/USD exchange was not that high. This changed following Donald Trump’s election victory. The spread widened significantly providing a clear signal for currency investors causing a significant strengthening of the dollar versus the euro and hence a tightening of the correlation between the spread and the exchange rate.

Another example is the huge correlation in 2016 between the relative performance of European banks versus the overall Eurozone index and the level of German Bund yields. Investors look at the slope of the yield curve as a proxy for bank profitability: banks do maturity transformation and higher bond yields will make credit also more expensive. As a consequence, when bond yields rise, bank share prices outperform the index.

One can wonder why correlations have become so high at times. One reason is that an increasing number of investors have become aware of them. A second is that ETFs have made it easy for many investors to put on correlation strategies. A third reason is that in a low rate and low expected return environment, once investors spot an opportunity, they play it to the full extent. For all of these reasons correlations may at times become very high.

There is a caveat however: correlations can break down, causing a rush towards the exit. Political decisions can be a factor underpinning changes in correlation. This was illustrated by the OPEC agreement to put a ceiling on production. Historically, the correlation between the effective dollar exchange rate and the price of oil has been negative. During Q4 2016, this correlation became positive: the dollar strengthened on the back of Donald Trump’s victory whereas oil rose following the OPEC agreement.

Even trickier is a situation when correlations can change signs. This one bears watching by equity investors in an environment of rising US interest rates. For many years now we have become used to a negative correlation between bond markets and equity markets: when the latter were going down, bond yields declined and the good performance of bond markets was cushioning the impact of negative equity performance on a diversified portfolio. At some point in a monetary policy tightening cycle, higher policy rates not only weigh on bond market performance but also on equity performance because of investor concern about the growth outlook: the correlation becomes positive and investors suffer twice, causing an increase in risk aversion and a rush towards cash.