It’s been said diversification is the only free lunch in investing, with the largest benefits of diversification coming from adding assets with low, or—even better—negative correlations.
However, when designing portfolios, investors also need to be aware that correlations are not constant—they are averages of the relationships of returns. Thus, it’s important to understand not only that correlations can drift, but also under what circumstances the correlation of returns are likely to increase and decrease.
For example, while the correlation of high-yield bonds to stocks tends to be low, during bear markets caused by recessions, their correlation tends to rise toward 1 (at exactly the wrong time).
Thus, high-yield bonds are not as effective as safer Treasury bonds (and other high-quality bonds; the lower the credit quality, the more equity-like the risk) at diversifying equity risks.
Demonstrating their nonstable relationship, while U.S. stock and bond return correlations have been predominantly positive over the last 40 years, there have been persistent episodes of large, negative stock/bond correlations. U.S. stock/bond correlations were about 0.2 during the 1970s; increased to, on average, 0.4 during the 1980s and first half of the 1990s; and fell to, on average, -0.2 since 1998.
Correlations also varied substantially over subperiods: They were as high as 0.7 in the fall of 1994, and as low as -0.7 in the second quarter of 2003 and third quarter of 2012. The evidence on stock/bond correlations in other developed markets followed a similar pattern, suggesting that one or more common factor(s) are driving international stock/bond correlations.
The shifting correlations can often be explained by demand and supply shocks. For example, a negative supply shock (such as the 1970s oil embargo) can lead to positive correlations between stocks and bonds, as can a positive supply shock (such as from fracking). On the other hand, negative demand shocks can explain much of the negative correlations observed since 2000. Demand shocks push inflation and output in the same direction, resulting in lower stock/bond correlations.
Correlations, Monetary Policy & Inflation
Lieven Baele and Frederiek Van Holle contribute to the literature with their October 2017 study, “Stock-Bond Correlations, Macroeconomic Regimes and Monetary Policy,” which examined the relationships between high, intermediate and low inflation, output regimes and monetary policy.
The authors note that, in terms of monetary policy, in an accommodating regime, output stabilization dominates policy, while in a restrictive regime, the central bank is mainly focused on controlling inflation. Their data set consisted of daily total equity index and 10-year benchmark bond returns for 10 developed markets (the United States, Canada, Japan, the U.K., Germany, France, the Netherlands, Belgium, Italy and Spain).
The dataset’s starting dates varied because of the lack of daily data, with the longest series being for the United States and starting in June 1961. For several other countries, however, the sample starts at the end of the 1980s or early 1990s. The end date is December 2013.
Following is a summary of their findings:
- There is a strong association between stock/bond correlations and monetary policy regimes.
- Stock/bond correlations tend to be slightly negative in recessions and positive in expansions.
- Negative stock/bond correlations are associated with periods of accommodating monetary policy, but only in times of low inflation.
- Irrespective of the inflation and/or growth regime, stock/bond correlations are always positive when monetary policy is restrictive.
- Pure inflation and growth regimes have little explanatory power for stock/bond correlations.
- In the low-inflation regime, inflation averaged 1.9%. Inflation volatility was also low, at 0.5%. In the intermediate-inflation regime, inflation averaged 4.2%, and its volatility was 0.9%. In the high-inflation regime, inflation averaged 10.5%, and its volatility was 2.7%. The low-inflation and high-inflation regimes both had an expected duration of about 11 years, compared to about five years for the medium-inflation regime.