Stock–bond correlation: when bonds protect stocks, and when they don't

For twenty years, from 2000 to 2020, bonds rose when stocks fell. Most portfolio advice written today assumes that is normal. Over 150 years it was the exception: stocks and bonds moved in the same direction in 58 % of all months since 1875. This guide explains what the stock–bond correlation is, why it changes, what actually protected portfolios in each regime, and how to check the regime you are in today.

What the stock–bond correlation measures

Correlation is a number between −1 and +1 that says how two things move together. At +1 they rise and fall in step; at −1 one rises when the other falls; at 0 there is no pattern. The stock–bond correlation compares the monthly returns of US stocks (the S&P 500, the 500 largest US companies) with the returns of US government bonds (Treasuries).

A negative correlation is what the classic 60/40 portfolio, 60 % stocks and 40 % bonds, relies on. When stocks fall, bonds are supposed to rise and cushion the fall. A positive correlation means the cushion is gone: in a bad month for stocks, bonds are more likely to fall too.

Because single months are noisy, the useful measure is the correlation over a window of years. The three-year (36-month) correlation is the one used throughout this guide and in our white paper.

150 years in three eras

Our White Paper No. 1, When Bonds Stop Hedging, measured the three-year stock–bond correlation back to 1875. Three eras stand out.

• 1875–1965: positive more often than not. Stocks and bonds spent 56 % of these months moving in the same direction, with no stable pattern, because there was no modern central bank for most of the period.
• 1966–1999: positive in 94 % of months. The inflation decades of the 1970s and 1980s and the long disinflation that followed kept stocks and bonds moving together. An investor who held bonds "for protection" in this era owned a second asset that fell in many of the same months as the first.
• 2000–2020: negative in 96 % of months. This is the era most of us learned investing in. Bonds rallied in the 2000–02 bear market, in 2008 and in March 2020. The 60/40 portfolio earned its reputation here.

Then came 2022. From its peak in December 2021 to the low in September 2022 the S&P 500 fell 24 %, and 10-year Treasuries fell 17 % over the same months. The correlation turned positive, and by December 2024 the three-year reading had reached +0.72, the highest in the full 150-year sample. In September 2026 it stood at +0.48.

The point is not that bonds are useless. The point is that the 2000–2020 behaviour was one regime among several, and regimes last years, not weeks.

Why the correlation changes: inflation

The white paper's clearest result is an inflation gradient. Since 1955, when US inflation was above 4 %, the following three years showed a positive stock–bond correlation in 94 % of months. When inflation was below 2 %, the correlation was positive only about a third of the time.

The mechanism is short to state. When inflation is the main worry, rising prices hurt both assets at once: bonds lose because their fixed coupons buy less, and stocks lose because interest rates rise to fight inflation. When growth is the main worry, a weak economy hurts stocks but helps bonds, because interest rates fall. Growth scares produce a negative correlation; inflation scares produce a positive one.

This is also why the two eras look the way they do. The 1966–1999 era was an inflation story; 2000–2020 was a low-inflation story with three growth scares. 2022 was an inflation shock, and bonds behaved exactly as the longer history suggests they would.

What this means for a 60/40 portfolio

A 60/40 portfolio is built on the assumption that its two halves offset each other. When the correlation is positive, the portfolio is riskier than its 2000–2020 history suggests, because the fall in a bad month is no longer cushioned. In 2022 a 60/40 mix fell further than many of its holders thought possible, not because either asset did something unprecedented, but because they did it together.

That does not make 60/40 dead. It makes it regime-dependent. The practical question is not "should I hold bonds?" but "which regime am I in, and what else can carry the defensive job?" Our guide to the 60/40 portfolio covers the construction side; here we stay with the evidence on what protected portfolios.

Twelve sell-offs: what protected a portfolio in each regime

Since 1955 the S&P 500 has fallen more than 15 % twelve times. The white paper sorted them by the three-year correlation at the market peak, and by what the defensive assets returned from that peak to the low:

Peak-to-trough, month-end total returns of each defensive asset (not of a whole portfolio). Bold = the better of Treasuries and cash. Source: White Paper No. 1 (Shiller, FRED, Yahoo Finance).

Five sell-offs with a clearly negative correlation going in: 1957, 1962, 2000–02, 2008–09 and 2020. In all five, Treasuries did what the textbook promises and gained between 2 % and 30 % while stocks fell.

The sell-offs with a positive correlation going in told a different story. In 1969–70 and 1973–74 Treasuries were roughly flat, while cash (Treasury bills) earned 13–14 %. In 1980–82 Treasuries returned +19 %, but cash returned +25 %. Across all sell-off months with a positive correlation, Treasuries beat cash only about one time in three.

2022 is the warning about slow signals. At the December 2021 peak the three-year reading was still −0.39, because the previous three years were dominated by the 2020 crash, when bonds hedged. The correlation then flipped inside the sell-off, and 10-year Treasuries lost 17 % while stocks lost 24 %. A three-year window is stable, but late.

Gold is often named as the answer. Its record is two-faced: +138 % in 1973–74, when the bond hedge was broken, but −46 % in 1980–82, when inflation ran above 10 %. Gold is the asset that did most in 1973–74; it is not a reliable inflation hedge. Our gold-vs-bonds guide goes deeper.

The lesson across all twelve: no single defensive asset worked in every regime. Bonds carried the growth scares; cash carried the inflation scares; gold helped in one crisis and hurt in the next.

Can you tell in real time?

If regimes matter this much, the obvious wish is a signal that tells you in advance when bonds will stop hedging. The white paper tested six real-time signals. The best of them improved on a naive guess by 5–6 percentage points. Two popular ones, rising inflation and the bond price trend, had no predictive value at all.

Two warnings follow from this. First, forecasting the correlation is hard; knowing the regime you are in is easy, because the three-year correlation is a published number. Second, a rule built on one indicator can backfire. In 2008 US inflation was high because of the oil spike. A rule that sold bonds when inflation was high would have sold them just before Treasuries rallied 17 % in the financial crisis.

The white paper also tested a rule that switched the bond sleeve into cash when the correlation turned positive. It improved the Sharpe ratio (return per unit of risk) from 0.51 to 0.54 over 70 years, beat 86–88 % of 2,000 random switching schedules, and drew most of its gain from 2022–2026. Backtested, hypothetical, and described in the paper as "probably small, not proven". Our guide on market timing explains why that is the right way to read such a test.

What held up: don't let one asset carry the defensive job

The most robust result in the white paper is not a timing rule. It is diversification of the defence. A static mix of 60 % stocks, 20 % Treasuries and 20 % gold had a Sharpe ratio of 0.56 over 1973–2026, against 0.51 for 60/40, without any switching. Backtested results are hypothetical.

The reason is the record above. Bonds and gold failed in different crises, so holding both meant one of them usually worked. Cash, when Treasury bills pay 4–5 %, is a real third option: it does not fall when interest rates rise.

The cost of this approach is that gold is volatile and can lose for a very long time: after inflation it lost more than half its value between 1980 and 2001. Size it so you can sit through a decade like that, not just the good years. Our guide on crash protection shows how the pieces fit together.

How to check the regime you are in

The three-year correlation is the one number that tells you whether your bonds are likely to protect you in the next sell-off. You can compute it yourself from monthly returns, or read it from our Hedge Health indicator in the free Research Terminal, which shows the current state as GREEN, AMBER or RED together with the 36-month and 12-month correlations and the latest inflation reading.

A reading that has been positive for years, combined with inflation above 4 %, is the regime in which bonds did least. A reading that has been negative, with inflation below 2 %, is the regime in which they did most.

A checklist for the defensive part of your portfolio

• Know your regime. If stocks and bonds have moved together over the last three years, do not count on bonds in the next sell-off.
• Don't let one asset carry the whole defensive job. A mix of Treasuries, cash and, optionally, gold was more robust than any single asset or any switching rule.
• Treat cash as a real asset when rates are high. At 4–5 % T-bill yields, cash does not fall when rates rise.
• Be careful with inflation-based rules. 2008 is the warning.
• Slow signals are stable, but late. At the December 2021 peak the three-year correlation was still −0.39. A rule based on it held bonds into 2022.
• Size by risk. In a positive-correlation regime a 60/40 is riskier than its recent history suggests.
• Re-check every quarter. Regimes last years, not weeks.

The full study, with all twelve sell-offs, the six signals and the tests that did not work, is our free white paper When Bonds Stop Hedging. The current Hedge Health reading is in the free Research Terminal; no credit card is needed.

Frequently asked questions

Hamann, H. (2026). Stock–bond correlation: when bonds protect stocks, and when they don't. InvestWealth.com. https://www.investwealth.com/learn/guides/stock-bond-correlation

<a href="https://www.investwealth.com/learn/guides/stock-bond-correlation">Stock–bond correlation: when bonds protect stocks, and when they don't</a> (InvestWealth.com)