The State of Wisconsin Investment Board is incorporating top-down macro analysis of the drivers of stock-bond correlations into its risk management, including to assess the potential of a secular shift in the stock-bond correlation.
The need to include analysis of the macro scenarios that drive a potential shift in the stock-bond correlation are highlighted in a paper co-authored by Edouard Senechal senior portfolio manager at State of Wisconsin Investment Board, recently published in the Financial Analysts Journal.
“As a result of the paper we are working on alternative risk analysis to better understand the macro influences in our portfolio,” Senechal told Top1000funds.com in an interview.
The paper, written in collaboration with researchers at Robeco, Empirical evidence on the stock bond correlation, shows that abrupt regime shifts in correlation can follow long periods of relative stability. And by examining data as far back as 1801 it shows that inflation, real rates and government creditworthiness are important explanatory variables of the stock bond correlation.
“The macro variable can be very stable for a long time and then can shift, that is a risk that needs to be assessed right now,” Senechal says. “Most risk models take a bottom-up lens looking at things like style and sectors . The characteristics of companies have been defining the way we look at risks. At the moment most risk models are based on bottom up data but there is a need to also act on top down macro analysis. We are changing the lens.”
Senechal points to data from the US that finds between 1970 and 1999 the average stock bond correlation was 0.35 and then was −0.29 between 2000 and 2023.
“I was at a macro conference and one of the participants made a joke about correlations. When he was asked what is your view on the level of stock bond correlation, the answer was simple. Everyone knows it’s 0.3, the only thing is to work out if it is positive or negative,” he says. “But jokes aside this is very important. For the last 30 years the correlation has been negative, but for the previous 30 years before that it was positive 0.35. This completely changes your asset allocation and policies.
“The correlation between stocks and bonds is the cornerstone of asset allocation but until recently it has received little attention because it doesn’t impact until there is a big shift.”
This has important implications for asset allocation and portfolio policies. Everything else equal, the difference between the correlations of 1970-1990 and 1999-2023 results in a 20 per cent increase in risk to a 60:40 portfolio, a corresponding drop of 20 per cent in the Sharpe ratio and a significant impact on returns.
“Most people use data from the last 20-30 years, but that is not necessarily reflective of what we will get in the next 20-30 years,” Senechal says.
The paper uses a large sample looking back to 1875 in the US and 1801 in the UK, and in examining the macro drivers.
“In the post 1950s environment central bank policies started to resemble those of the present day with a dual mandate and the objective of managing both inflation and unemployment.
“When inflation is low, as it was over the last 30 years, then they set nominal rates mainly as a function of unemployment. During downturn as in 2000 or 2008 or 2020, they cut rates and enter the QE program, when equities are selling off.
“This creates a negative correlation between stocks and bonds, which makes bonds’ hedging characteristics extremely attractive to investors.
“Therefore, inflation and real rates level are important determinants of the correlation. The question today is: are we facing a structural change in inflation after 30 years of decline. Understanding when these long-term trends change, or break is critical.”
Understanding inflation and AA implications
Senechal says the key variable right now is inflation. Referencing a Top1000funds.com interview with chief strategist at IMCO, Nich Chamie he says a structural change in globalisation could result in higher inflation.
“The rise in inflation due to COVID is disappearing now but that doesn’t mean that underneath the peak from COVID there isn’t a new problem that is caused by the decline in globalisation. The question is if we are in this environment, as IMCO says, inflation could be structurally higher and that will mean an environment of higher stock bond correlation, which will have a big impact on the risk of a diversified portfolio.”
The team at SWIB is working through the implications for the portfolio and any asset allocation shifts that may need to occur. It has already reduced leverage from 15 to 12 per cent, reflecting rates going up over the past three years, and if that continues leverage is less important and will be reduced further.
“We are doing a lot of work on developing risk models that allow us to measure what type of sensitivity we have to real rates, inflation and growth,” he says. “If the stock bond correlation keeps going in the same direction, being positive, then probably returns on bonds will decline as investors ask for higher bond risk premia and therefore higher yields.”
This is a good paper for explaining contemporaneous stock bond correlation. As the SWIB paper mentions, the prospect of predicting the stock bond correlation is left for future research. We have a paper that addresses prediction (3 month forward horizon), under review by the FAJ and presented in the July 2023 JPM Quant Conference. It is on SSRN, called “A Changing Stock-Bond Correlation: Explaining Short-Term Fluctuations”.