Correlation

Quick Definition

In 2022, both stocks and bonds fell at the same time, and the 60/40 portfolio lost 17%. The reason was correlation. Correlation measures how two assets move in relation to each other, ranging from -1 (they always move in opposite directions) to +1 (they always move together). A correlation of 0 means no relationship. This single number determines whether your diversification strategy actually protects you or just gives you a false sense of safety.

What It Means

Correlation is the statistical backbone of portfolio construction. When you hold assets with low or negative correlation, a loss in one tends to coincide with a gain or smaller loss in the other. That is the entire point of diversification: spreading risk across assets that do not all decline simultaneously.

Harry Markowitz's Modern Portfolio Theory (1952) demonstrated mathematically that you can construct portfolios with better risk-adjusted returns by combining assets with low or negative correlations, even if each individual asset is risky in isolation. The lower the correlation between assets in your portfolio, the more risk you can reduce without sacrificing expected returns.

The problem is that correlation is not fixed. It changes with the economic regime. From 1998 through 2021, the stock-bond correlation was persistently negative, typically running between -0.20 and -0.40. When stocks fell, bonds tended to rally, cushioning the blow. That is what made the 60/40 portfolio work for two decades.

Then 2022 happened. Core CPI ran above 6%, the Federal Reserve raised rates from near 0% to 5.25% in under two years, and both stocks and bonds sold off together. The rolling 24-month correlation between S&P 500 daily returns and 10-year Treasury total returns peaked near +0.50 in mid-2023. The 60/40 posted its worst calendar year since 1937.

As of early 2026, the rolling 24-month stock-bond correlation has drifted down from its 2023 peak but has not returned to the deeply negative levels of the 1998-2021 era. It appears to be running in the +0.10 to +0.30 range, depending on the window and bond proxy used. The 12-month correlation dropped to approximately 0.16 by late 2025, according to State Street research. This transitional zone makes asset allocation decisions more difficult than they were when you could simply assume bonds would cushion equity losses.

How It Works

Interpreting Correlation Coefficients

The Diversification Math

Two assets, each with 15% annual standard deviation (volatility):

By combining two equally risky assets with zero correlation (50/50), portfolio volatility drops from 15% to 10.6%, a 29% reduction with no expected return sacrifice.

Real-World Examples

Example 1: The 2022 Correlation Break

The 60/40 portfolio lost roughly 17% in 2022, its worst calendar year since 1937. The S&P 500 dropped about 19% while the Bloomberg US Aggregate Bond Index fell about 13%, its worst year on record. For the first time since 1969, stocks and bonds both posted negative returns in the same calendar year. The stock-bond correlation spiked to approximately +0.50, the highest reading in the modern sample. Equities and bonds declined together for 14 consecutive months.

The mechanism was straightforward. Rising interest rates crushed bond prices while simultaneously compressing equity valuations. Both assets were repriced by the same lever. There was no shock absorber because the shock and the absorber were the same thing.

Example 2: The Recovery (2023-2025)

The 60/40 was not dead. It returned approximately 17% in 2023, 15% in 2024, and 15% in 2025. Three consecutive years of double-digit returns, roughly double the long-term average of 7.8%. The portfolio that was declared dead in late 2022 compounded at roughly 15% annually over the three years since.

The correlation story drove the recovery. The 12-month stock-bond correlation dropped from its peak of 0.80 in mid-2024 to just 0.16 by late 2025. The macro regime shifted from "inflation is the only story" (which drives stocks and bonds together) to "growth versus inflation" (which allows them to diverge again). When growth concerns dominate, stocks fall but bonds rally as investors flee to safety. That is the normal dynamic that makes 60/40 work.

Example 3: The 2026 Iran Crisis and Crisis Correlation Convergence

In March 2026, a geopolitical energy shock involving Iran caused a notable further deterioration in cross-asset correlations. During this 15-trading-day window, gold declined alongside equities, likely reflecting a tighter anticipated monetary policy response than in prior energy crises. This was notable because gold had historically been the most effective hedge during 1970s stagflationary shocks. The stock-bond correlation turned positive again (approximately +0.47), resembling the 2022 regime.

This crisis reinforced a well-documented pattern: in deflationary crises (2008, 2020), the negative stock-bond correlation preserves 60/40 diversification. In inflationary crises (1973, 1979, 2022, 2026), this correlation turns positive, degrading conventional balanced construction.

Historical Correlations Between Major Asset Classes

Why Stock-Bond Correlations Change Over Time

The most important correlation in portfolio construction, stocks versus bonds, is not stable. The empirical cutoff sits around a 2.5% core inflation rate. Below that, correlation tends to be negative. Above that, correlation tends to be positive or near zero.

The BEA's monthly core PCE series is the cleanest read on where we are. Recent monthly prints have run in the 2.6 to 2.8% range as of early 2026, right at the inflection zone. If core inflation runs sticky at 2.5 to 3%, the correlation likely stays near zero or mildly positive for the duration of the regime.

Rolling Correlation: How It Changes Over Time

Correlations are not static. They evolve with market regimes.

US Stocks vs. Bonds rolling correlation (illustrative):

• 1995-2000: +0.30 (positive, inflationary era ending)
• 2002-2008: -0.25 (negative, deflation/low inflation era)
• 2010-2019: -0.30 (low inflation; bonds reliably hedged stocks)
• 2022: +0.50 (inflation spike broke the negative relationship)
• 2023: Declining from peak
• 2025: 0.16 (12-month, per State Street)
• Early 2026: +0.10 to +0.30 (24-month, transitional band)

Using only long-term average correlations misses the regime-dependency that can cause diversified portfolios to fail in specific environments.

Crisis Correlation: The Problem With Tail Risk

In market crises, correlations between risk assets tend to converge toward +1.0.

March 2020 (COVID crash) peak stress correlations:

• US stocks vs. international stocks: ~+0.95 (near perfect sync)
• US stocks vs. corporate bonds: ~+0.85 (both sold off)
• US stocks vs. REITs: ~+0.90
• US stocks vs. commodities (oil): ~+0.85

Almost everything fell together. The only reliable crisis diversifiers were US Treasuries, cash, and gold (though gold briefly failed during the 2026 Iran crisis as well).

This correlation convergence is a well-documented phenomenon: diversification benefits disappear precisely when you need them most. This is why truly defensive portfolios hold genuine safe havens (short-term Treasuries, cash), not just equity diversification. Some institutional investors have responded by adding explicit convexity strategies (options, trend-following) as regime-independent portfolio protection.

Key Points to Remember

• Correlation ranges from -1.0 to +1.0; lower correlations produce greater diversification benefits
• The stock-bond correlation is the most important in portfolio construction, but it is not stable and failed dramatically in 2022
• The empirical cutoff for stock-bond correlation is around 2.5% core PCE inflation; below that, correlation is negative; above that, it turns positive
• As of early 2026, the 24-month stock-bond correlation is in the +0.10 to +0.30 range, a transitional band between the 2022-2023 highs and the 1998-2021 negative era
• Near-zero correlation assets (gold, commodities) provide the best diversification alongside stocks, though even gold can fail in inflationary crises
• Combining two assets with zero correlation reduces portfolio volatility by approximately 29% with no expected return sacrifice
• Rolling correlations reveal how relationships shift over time; static averages miss important regime changes

Related Concepts

• Diversification: The strategy of spreading investments across assets with low correlation
• Portfolio: The complete collection of investments managed together
• Asset Allocation: How you divide investments across asset classes; correlation determines optimal weights
• Risk Management: Identifying and mitigating financial risks, including correlation risk
• Beta: Measures a stock's volatility relative to the market; related to but distinct from correlation
• Bond: Fixed-income instrument; the bond side of the 60/40 portfolio
• Federal Reserve: Central bank whose rate decisions drive the inflation regime that determines stock-bond correlation
• Inflation: The variable that determines whether stock-bond correlation is negative or positive

Common Mistakes to Avoid

• Assuming correlation is stable: The stock-bond correlation was negative for two decades, then flipped positive in 2022. Using only long-term averages in portfolio construction hides regime risk.
• Believing diversification works in all crises: In deflationary crises (2008, 2020), bonds hedge stocks. In inflationary crises (2022, 2026), both fall together. Know which type of crisis you are in.
• Overestimating gold as a safe haven: Gold diversifies well in normal times and deflationary crises, but during the 2026 Iran energy shock, gold declined alongside equities.
• Ignoring the inflation cutoff: The empirical cutoff for stock-bond correlation is around 2.5% core PCE. With core PCE at 2.6 to 2.8% in early 2026, we are right at the inflection zone. Portfolio decisions should account for the possibility that correlation stays positive.
• Using levered risk parity without stress-testing correlation assumptions: Levered risk-parity strategies assume negative stock-bond correlation in their sizing. When the correlation flipped positive in 2022, the leverage amplified rather than cushioned drawdowns.

Frequently Asked Questions

Q: If international stocks have high correlation with US stocks, why own both? A: Correlation of 0.85 still provides some diversification benefit because it is not perfect correlation. Additionally, over very long periods, international stocks can outperform US stocks significantly. The 2000s were a US underperformance decade when international stocks vastly outperformed. Expected return differences are a reason to diversify internationally beyond just the correlation argument.

Q: Is Bitcoin a good diversifier? A: Historically, Bitcoin's correlation with US stocks has been low over long periods, suggesting diversification value. However, during acute risk-off events (COVID crash of March 2020, 2022 bear market), Bitcoin sold off aggressively alongside stocks, exhibiting crisis correlation convergence. The diversification benefit is real but unreliable during exactly the periods when you need it most.

Q: How do I calculate correlation between two assets? A: Correlation (r) = Covariance(A,B) / (Standard deviation of A x Standard deviation of B). In Excel: =CORREL(Array1, Array2). Most investment platforms and portfolio analysis tools display rolling and historical correlations automatically. For personal portfolio analysis, tools like Portfolio Visualizer (portfoliovisualizer.com) provide free correlation matrices.

Q: Will the stock-bond correlation return to negative? A: It depends on inflation. If core PCE decisively falls below 2.5%, the correlation should drift back negative, restoring the traditional 60/40 diversification benefit. If core inflation runs sticky at 2.5 to 3%, the correlation likely stays near zero or mildly positive. The 60/40 has still delivered strong returns (17% in 2023, 15% in 2024, 15% in 2025) even with elevated correlation, because real bond yields above 1.8% provide genuine income. See this FRED series for the current 10-year Treasury yield.