Asset Class Correlation

Hi all—I hope everyone is having a great summer and had a wonderful Fourth of July!

I wanted to dive deeper into a topic this month and periodically as we continue to build on our monthly TLG Advisor-focused newsletter. This month, I’ll dive into asset class correlation.

Diversification

Everyone knows that it can be risky to put all of your eggs in one basket. For most people, it is prudent to make sure you spread out investments. Nobel prize winner Harry Markowitz has been quoted as calling diversification, “the only free lunch in finance.” Diversification typically improves risk-adjusted returns, therefore increasing the amount of return for every unit of risk. One of the main drivers of the risk reduction inherent in diversification is the correlation among investments within a given portfolio. The risk, measured by variance, decreases (all else being equal) with a decrease in the correlation coefficient.

  • Portfolio variance[1] = w12σ1+ w22σ2+ 2w1w2Cov1,2

Where:

  • w1 = the portfolio weight of the first asset
  • w2 = the portfolio weight of the second asset
  • σ= the standard deviation of the first asset
  • σ2 = the standard deviation of the second asset
  • Cov1,2 = the co-variance of the two assets, which can thus be expressed as p(1,2)σ1σ2, where p(1,2) is the correlation coefficient between the two assets

Correlation

Correlation is how things move together, and correlation can range from -1 to 1. It’s easy to think of it as a tri-polar measure. -1 would suggest they are inversely correlated, i.e. bond prices and interest rates. Zero would indicate there isn’t much statistical relationship between the movement of two things. There are easy examples to think of here – the temperature in the room you’re sitting in doesn’t have much correlation with today’s index performance for instance. A correlation of one would be perfect correlation; how tall you are in centimeters is perfectly correlated with your height on the imperial system of measurement. We often hear the phrase, “Correlation does not equal causation.” Intellectually we know this, but at the same time, instinctively we often see correlation and causation moving hand-in-hand as we are wired to find patterns. Often a third thing is more likely a causal factor. Ice cream sales and beach vacations may have strong positive correlations, but clearly hotter weather is the catalyst of both.

When it comes to asset class correlation, not many things sit on these poles for long if ever. Correlation tends to be unstable for a variety of factors. When you look at simply a U.S. stock-bond correlation, it depends upon a myriad of factors, including real interest rates and inflation and the uncertainties around both. In the period between 2000 and 2023, we had relatively lower interest rates and inflation which contributed to lower correlation than in years past.[2] In the 30 years prior, there were higher correlations between the stock and bond asset classes, more inflation, and higher interest rates.

After a few years of higher-than-usual inflation and an aggressive interest rate hiking cycle, we will likely see these correlations shift higher based on historical data and trends, which can change the way we look at risk to portfolios. This is of course not guaranteed as markets continually prove that they are much more complicated than a spreadsheet with data analysis can make it seem. For example, a JP Morgan article published late last year suggested that both bonds and equities faced interest rate sensitivity in 2022, boosting correlations. The article suggests that these correlations may normalize (and fall back below zero at least temporarily) once the Fed pivots to a less restrictive policy stance.[3]

Takeaways

Correlations are unstable, and portfolio risk-adjusted returns favor lower correlation asset classes. Correlations between stocks and bonds can change significantly across different interest rate and inflation regimes. The lower correlations we’ve seen so far this century between stocks and bonds are not guaranteed to last indefinitely. The 60-40 portfolio introduces more risk all things being equal as correlations rise, so it is important to stay aware and understand how insurance and other diversifying strategies[4] can help keep risk in line with consumer expectations.


[1] Hayes, A. (2024, May 28). Portfolio variance: Definition, formula, calculation, and example. Investopedia. https://www.investopedia.com/terms/p/portfolio-variance.asp

[2] Swinkels, L. (2024, June 17). New research into the stock-bond correlation shows when they correlate – and when they don’t: Robeco USA. Robeco.com – The investment engineers. https://www.robeco.com/en-us/insights/2024/05/new-research-into-the-stock-bond-correlation-shows-when-they-correlate-and-when-they-don-t

[3] Manley, J. (2023, September 15). What is the outlook for the relationship between stocks and bonds?. What is the outlook for the relationship between stocks and bonds? | J.P. Morgan Asset Management. https://am.jpmorgan.com/us/en/asset-management/protected/institutional/insights/market-insights/market-updates/on-the-minds-of-investors/what-is-the-outlook-for-the-relationship-between-stocks-and-bonds/

[4] A changing stock-bond correlation. AQR Capital Management. (n.d.). https://www.aqr.com/insights/research/journal-article/a-changing-stock-bond-correlation

Ben Tiller

Director of Advisory Services