Cross-Asset Correlation: How Asset Classes Move Together (and When They Don't)

Diversification is the closest thing to a free lunch in investing — the idea that by combining assets that don't all fall simultaneously, a portfolio can achieve better risk-adjusted returns than any single asset. But diversification is only as reliable as the correlations between assets, and correlations are neither fixed nor stable. In market crises, they frequently converge toward one: assets fall together at precisely the moment that diversification is most needed.

Understanding correlation — what drives it, when it breaks down, and which asset classes genuinely offer low or negative correlation to equities — is fundamental to building portfolios that hold up across the full range of economic conditions.

What Correlation Means

Correlation is a statistical measure ranging from -1 to +1:

• +1.0: Two assets move in perfect lockstep — when one rises 10%, the other rises exactly 10%.
• 0: No linear relationship between the two assets' movements.
• -1.0: Perfect inverse relationship — when one rises 10%, the other falls 10%.

In portfolio construction, the goal is not to maximise return from each individual asset but to build a combination whose aggregate volatility is lower than the weighted average volatility of its components. This reduction in portfolio volatility is only possible because of less-than-perfect correlation between positions.

The mathematics is clear: a portfolio of two assets with correlation +1.0 achieves no volatility reduction. A portfolio of two assets with correlation -1.0 can theoretically eliminate volatility entirely (though with zero expected return). Real portfolios operate between these extremes.

Historical Correlation Patterns

Equities and government bonds

The most important cross-asset relationship in multi-asset portfolio theory is between equities and high-quality government bonds (gilts, US Treasuries, German Bunds). This correlation has oscillated substantially over the past century:

1970–2000: Positive correlation — both equities and bonds were driven by inflation expectations. High inflation hurt bonds (rising yields) and hurt equities (higher discount rates). They moved together.

2000–2021: Negative correlation — the "deflation era." Recessions drove investors into safe-haven bonds (prices rose, yields fell) at the same time equities fell. The 60/40 portfolio worked exceptionally well in this period because bonds acted as a genuine hedge.

2022: Correlation turned sharply positive again. Rising inflation caused central banks to raise rates aggressively; bond prices fell sharply while equities also fell. The 60/40 portfolio had its worst year since the 1970s. The UK gilt market experienced extraordinary volatility following the September 2022 mini-Budget.

The lesson is that the equity-bond relationship depends heavily on the macroeconomic regime — specifically, whether inflation or growth is the dominant risk factor. When inflation is the primary concern, equities and bonds move together (bad for diversification). When recession/deflation is the primary concern, they diverge (good for diversification).

Equities and property

Listed real estate (REITs) has historically shown moderate positive correlation to equities — approximately 0.5–0.7 over the long run. In the short run, particularly during acute equity market sell-offs, REITs can fall in line with equities because they are traded on stock exchanges and subject to the same liquidity dynamics.

Direct property has lower correlation to equities — partly because it is valued infrequently (quarterly or annually) rather than continuously, which smooths the apparent correlation. Institutional investors should be aware that this "smoothed" correlation is partly an illusion of infrequent valuation rather than true economic independence.

Equities and gold

Gold has maintained a low and frequently negative correlation with equities over long periods, particularly during equity bear markets. In the 2008–2009 financial crisis, gold was roughly flat while equities fell 40–50%. In 2020, gold rose sharply as equities initially fell. The mechanism is intuitive: gold is a safe-haven asset that investors buy when they lose confidence in financial assets more broadly.

However, gold can also fall with equities in liquidity crises (as briefly in March 2020) when investors sell everything to raise cash. And gold underperforms significantly during long equity bull markets. The negative correlation is not consistent; it is conditional on the nature of the crisis.

Equities and commodities

Broad commodity indices have shown low correlation with equities historically. However, this varies enormously by commodity:

• Oil: Correlated with equities in growth-driven environments (both rise as the economy expands). Negatively correlated in supply-shock environments (oil price spike hurts equities via cost pressures).
• Agricultural commodities: Low correlation with equities — driven mainly by weather, crop cycles, and geopolitical supply factors.
• Copper: Positively correlated with global economic growth; tends to fall with equities in recessions.

Equities and alternatives

• Hedge funds (long/short equity): Moderate positive correlation — despite the "hedge," net long positions mean most long/short equity funds fall with equities in bear markets.
• Managed futures (trend following): Consistently low or negative correlation in equity bear markets. Trend-following strategies shorted equities in 2008 and 2022, generating positive returns as stocks fell.
• Private equity: High long-run correlation with public equities, but appears lower due to quarterly valuations and J-curve dynamics.
• Infrastructure: Lower correlation with equities; revenues are typically contractual and inflation-linked, less sensitive to the economic cycle.

Correlation in Crises: The Contagion Problem

The most dangerous feature of correlation for portfolio investors is contagion: during severe market stress, correlations between risky assets rise sharply and simultaneously. Assets that were 0.3 correlated during calm periods become 0.7 or 0.9 correlated when investors are forced to sell across the board.

The mechanism is largely institutional: when levered investors face margin calls or redemptions, they sell whatever they can, regardless of fundamental relationships. Forced selling creates artificial correlation where there is no economic connection.

This dynamic was observed in:

• 2008: Equities, credit, property, commodities, and emerging markets all fell simultaneously. Only government bonds and gold provided meaningful protection.
• March 2020: In the initial liquidity panic, even gold and US Treasuries fell briefly as investors liquidated everything.
• 2022: Both equities and bonds fell, leaving multi-asset investors with nowhere to hide among traditional asset classes.

The implication is that diversification is most reliable in normal markets and least reliable in the markets where it is most needed. True crisis protection requires assets that are fundamentally, not just statistically, uncorrelated with equities — meaning their returns are driven by entirely different economic mechanisms.

Genuinely Low-Correlation Assets

Short-dated government bonds and cash: In an equity sell-off driven by recession fears (not inflation), short-dated gilts will be stable or rise. Cash always has correlation zero with any declining asset.

Trend-following managed futures: AHL (Man Group), Winton, Millburn — these strategies have a documented history of positive returns during equity bear markets because they systematically follow price trends in bonds, currencies, and commodities.

Insurance-linked securities (catastrophe bonds): Cat bond returns are driven by natural disasters, not financial markets. Correlation with equities is close to zero in normal conditions.

Litigation finance: Returns linked to legal case outcomes — entirely uncorrelated with equity markets.

Royalty streams: Music royalties, pharmaceutical royalties — driven by usage/sales patterns, not markets.

Physical gold: Conditional negative correlation with equities; most consistent in deflation/financial crisis scenarios.

Practical Implications for Portfolio Construction

Avoid over-relying on historical correlation

Correlation matrices in portfolio optimisation software are based on historical data and will understate crisis correlation. The 2022 experience has prompted many advisers to reduce reliance on corporate bonds as an equity hedge and to incorporate more genuine non-correlated alternatives.

Distinguish economic and statistical independence

An asset that appears uncorrelated with equities because its valuations are infrequent (private equity, direct property) is not the same as an asset that is economically independent from equities. When the economic cycle turns, both will fall — just with different timing.

Build correlation-aware layers

A well-structured multi-asset portfolio might include:

• Tier 1 (growth engine): 50–60% equities — global, factor-diversified
• Tier 2 (traditional hedge): 15–20% short-dated gilts and cash — negative correlation in recession
• Tier 3 (genuine alternatives): 15–25% assets with low economic correlation — trend following, gold, infrastructure, insurance-linked

Be honest about bonds in an inflationary environment

If the primary macro risk is persistent above-target inflation rather than recession, long-dated government bonds are a poor equity hedge. In that environment, inflation-linked gilts, commodities, infrastructure, and real assets deserve larger allocations.

Monitoring and Reviewing Correlations

Correlation is not a set-and-forget parameter. Investors should review cross-asset correlation in their portfolios at least annually, and more frequently following significant macro regime changes. Key questions:

• Has the equity-bond correlation changed sign? (Check against the current inflation regime.)
• Are alternative allocations still delivering the uncorrelated return expected?
• Have any "alternative" positions become more equity-correlated as the underlying holdings have evolved?

Portfolio stress testing — simulating the portfolio under scenarios such as a 30% equity fall, a 200-basis-point gilt yield rise, or a dollar-sterling currency shock — provides a more rigorous check on real portfolio diversification than historical correlation tables alone.

Compliance Note

Correlation analysis is based on historical data which may not reflect future behaviour. Past correlations between asset classes can change significantly due to changes in monetary policy, inflation regimes, and market structure. Alternative and illiquid investments carry additional risks not captured by correlation statistics. This guide is for educational purposes and does not constitute personal financial advice. Investors should consult a qualified adviser before making any allocation decisions.

How Global Investments Can Help

Global Investments builds cross-asset portfolios for internationally mobile HNW clients using forward-looking analysis of economic regimes — not purely backward-looking historical correlations. We assess how your current allocation behaves under stress scenarios, identify genuine diversifiers appropriate to your liquidity needs and risk tolerance, and design portfolio structures that hold up across a range of macroeconomic outcomes. Contact our team to discuss a portfolio review.