Why rising stock-bond correlations matter

By Zac Tate, Hottinger Investment Management

 

There has been a rise in interest in what one might ordinarily consider to be a mundane subject: the correlation between stock market returns and bond returns. It is an important topic because portfolio asset allocation since the turn of the century has been built using assumptions about the correlation between market movements in stocks and bonds that are beginning to look a bit weaker. If these assumptions break down, then the consequences for investment strategy and performance could be profound.

First, we must be clear on the concepts because it is easy to get confused. Analysts often use terms such as prices, returns and yields inconsistently when talking about correlations between stocks and bonds. When thinking about the subject, one should therefore bear the following in mind:

  1. The price of a fixed coupon bond is always negatively correlated with the yield of the bond. A more expensive bond “yields” less as the interest payments (or coupons) are usually fixed against, in this example, the higher cost of purchasing the bond. Exceptions may apply to inflation-linked bonds, where higher inflation could raise coupons and therefore yields in a way that attracts demand from investors that pushes bond prices higher.
  2. The total return of a fixed coupon bond is always positively correlated with the price of the bond. The total return of a bond comprises both interest income received from the bond and any gains or losses in the capital value of the bond, measured by its price. Interest income is often – and confusingly – referred to as yield, but – as explained above – this is usually a fixed amount determined in advance of purchase.
  3. The total return of an equity comprises both dividend income and any gains or losses in the capital value of the equity, again measured by its price. The total return of an equity is usually positively correlated with the price of that equity unless companies explicitly tie their dividends to share price performance.

So if, for example, an analyst says that stock prices are negatively correlated with bond yields, they are almost certainly also saying that (i) stock prices are positively correlated with bond prices and (ii) total stock returns are positively correlated with total bond returns.  

Analysts have been studying the long-run stock/bond correlation data and believe that an important recent trend is coming to an end.

As Figure 1 shows, during most of the period since 1937 in the United States, the correlation in total returns has been positive, meaning that neither asset class offers protection from poor performance within the other. However, since the late 1990s, the correlation between total returns of stocks and bonds has been sharply negative, supporting a popular portfolio allocation model that puts roughly 60% of capital in equities or stocks and 40% in bonds. This is done with the expectation that when one asset class performs badly, the losses can be capped by stronger performance in the other asset class.

Alternative data series show that strong positive correlations in total returns were the norm not just in the U.S. but also in developed economies going back to the 1800s before turning negative in the 1990s. However, this era of negative correlations may be coming to an end everywhere.

Figure 1 shows the correlation in annualised total returns of the U.S. S&P 500 Stock Market (Equity) and 10-Year US Treasures (Bonds) on a rolling 10-year historical basis.

Our view, shared by others in the investment community, is that the period of negative total returns correlation has a proximate cause in low inflation, which the US has experienced since the late 1990s. We think the causes of low inflation are central bank independence – which only really took hold in the late 1990s and early 2000s – and weak bargaining power of labour, which is a trend that goes back to the 1970s. The share of national income in the US (and other developed economies) to wages has fallen from about 52% in 1970 to about 43% today, according to the Bureau of Economic Analysis.

In the last 20 years, monetary policy has prioritised low inflation and has typically taken the lead over fiscal policy when economies slow down or move into recession. It makes sense, therefore, for bond prices to rise as stock prices fall in the period before a recession as central banks prepare to cut interest rates. And it makes sense for this to unwind as economies recover. With anti-inflation central banks keeping a lid on demand and fewer opportunities for labour to push for higher wages, inflation has recently barely registered as a factor compared to how it has done throughout most of the rest of post-WW1 history.

Take, for example, the 1960s, the only period on historical record during which the correlation flipped from negative to positive. CPI inflation rose from around 1.2% in 1964 to 5.5% in 1969. According to Federal Reserve Economic Data, over the whole five-year period between 1965 and 1969, US 10-year Treasuries returned just 0.06%, a strongly negative real return given the high inflation. The S&P 500 returned 27%, which was not significantly more than the cumulative inflation – of 19.7% – over the period. Inflation may have affected both asset classes negatively; in the previous five-year period between 1959 and 1963, the S&P 500 returned 59.2% and US Treasuries rose by 19.2%. As both bonds and equities struggled together, the correlation between the two became positive and the historical 10-year correlation as shown in Figure 1 steadily rose. 

This may have been a singular case, but it is worth investigating whether there are any lessons for today.

Populist demands for a higher share of national income to be reallocated from capital to labour could create a structural shift which, combined with inflationary pressures, would push down returns across asset classes as higher labour costs cut into companies’ margins.  This appears to be what happened in the 1960s. Unemployment was below 4% – as it is today – which is considered a level that represents a tight labour market and raises the risk of wage-inflation.

Figure 2 shows the change in the share of U.S. national income or GDP paid in wages to labourers over a three-year historical period against the U.S. unemployment rate.

This is a view developed in some detail in a recent paper by Marco Pericoli for the Bank of Italy1. In it, Pericoli claims that before the 1990s, inflation and economic growth were negatively correlated as it was typically supply shocks that pushed up costs and brought about recessions. He found that in the stagflation period of the 1970s to 1990s, “good” news regarding future cash flows to equity was correlated with lower expected future inflation; in other words, supply factors such as labour or input costs were driving inflation higher and hitting company margins. Since 2000, however, inflation and economic growth have been positively correlated as demand shocks have driven the business cycle.   

Equally, populist demands for more active fiscal policies could generate higher wage inflation during stress periods for the economy. If wage inflation rises in response to a fiscal stimulus, stocks could fall as costs rise; bonds would be hit by higher price inflation. Therefore, total returns for bonds would tend to move in the same direction as total equity returns.  With increasing demands for fiscal stimulus at a time when unemployment is at record lows, it is not surprising that the stock/bond total return correlation is moving back towards positive territory.

The implications for asset allocation are potentially significant. If total returns between stocks and bonds stop offsetting each other, then the justification for running portfolios around the 60/40 model of roughly equal weightings to equities and bonds will start to fall away. Attention then ought to turn to judging each asset class on its own merits and allocating capital according to attitudes to risk and tolerance for drawdown. This may well be good news for active asset allocation and bad news for those who are simply resting on their laurels.  

1The paper also shows that the trends for historical stock-bond correlations apply for other developed economies such as the UK, France and Germany.