Why do different investments fall at the same time? 20 years of falling months

Investments that share a driver fall together, and the data shows it is a matter of drivers rather than labels. In the 24 worst months for world equities from January 2006 to April 2026, emerging-market and European equities fell every time, euro high-yield bonds 22 times and commodities 21 times.

Gold fell in 7 of those months and EUR-hedged global government bonds in 8, but the bonds' record weakened after 2021. For a household that may need to sell during a fall, the useful question is not how many funds it owns, but which of them have historically held their value in the months its equities lost most.

Key takeaways

1. In world equities' worst months, almost everything with growth risk fell too. On average 9 of the 12 asset classes we track lost value in MSCI World's 24 worst months, which each lost 4.1% or more.
2. For emerging markets and high yield, the link was clearly tighter in falling months than a constant relationship would produce. Emerging-market equities had a correlation of 0.72 with MSCI World in its falling months, against 0.61 expected if the relationship had been constant; for commodities, small caps and inflation-linked bonds the gap was 0.04 to 0.06.
3. Gold and high-quality government bonds were the exceptions, with conditions. Gold gained 2.1% on average in the 24 worst months; government bonds gained 0.5%, but lost 1.34% on average in falling equity months from 2022.
4. A low average correlation is not a promise for the next fall. It is an observation over one period and one inflation regime, and it describes direction more reliably than size.

What does correlation actually measure?

Correlation measures how consistently two return series move in the same direction, on a scale from -1 to +1. At +1 they rise and fall together; at 0 there is no linear relationship; at -1 they move in opposite directions.

Three things matter when reading one:

• The period. A correlation measured from 2006 to 2021 can differ sharply from one measured after 2022.
• The frequency. Monthly returns can tell a different story from daily or yearly ones; our horizon study shows how.
• The size of moves. A correlation of 0.3 says little about whether an asset loses 1% or 20% when equities fall.

Our diversification guide publishes the full correlation matrix. This study asks a narrower question: what happened in the months that hurt.

How we measured falling together

We used monthly total returns in euros for MSCI World and 12 asset classes, from January 2006 to April 2026, 244 months in all. MSCI World fell in 87 of them and rose in 157. For each asset class, we calculated:

• its correlation with MSCI World over all months, falling months and rising months;
• how many of MSCI World's 24 worst months it also fell in, and its average return in them.

There is a statistical trap here. Picking only volatile months tends to raise measured correlation, and picking only one side of the distribution tends to lower it, even when nothing about the relationship has changed. To separate a real change from that effect, we also calculated the correlation one would expect in falling months if returns were normally distributed with the full-period correlation, following Boyer, Gibson and Loretan (1999).

Correlations in falling months

Source: Copio calculation on index levels in euros; data by asset class (CSV).

Two patterns stand out.

Growth-sensitive assets were more tightly linked to equities when equities fell than when they rose. Emerging-market equities' correlation was 0.72 in falling months and 0.48 in rising ones, and 0.11 above what a constant relationship predicts. Euro high yield showed the same pattern: 0.67 when equities fell, 0.46 when they rose, 0.12 above the constant-correlation value. That fits a simple mechanism: in a sell-off, default risk, earnings risk and investors' appetite for risk deteriorate at the same time, so assets priced on those risks move as one.

High-quality government bonds, gold and cash showed no such tightening over the full period. Their observed values sit close to the expected ones, within 0.09. Euro corporate bonds sit between the two groups: a moderate correlation, without the extra tightening.

This matches what academic work has found for equity markets. Longin and Solnik (2001) report that correlation between international equity markets increases in bear markets but not in bull markets.

What fell in the 24 worst months

Correlation describes direction across many months. Counting the bad months directly is closer to the experience of an investor watching a portfolio.

Source: Copio calculation; month-by-month returns in the 24 worst months (CSV).

MSCI World lost 7.1% on average in these 24 months, from -13.1% in March 2020 to -4.1% in April 2025. In those months:

• Every equity region we track fell in at least 23 of 24 months. Owning US, European, Japanese and emerging-market equities spread the risk of one region lagging for years; it did not spread the risk of a bad month.
• Euro high yield fell 22 times and lost 3.5% on average. It behaved like a smaller dose of equity risk.
• Gold rose in 17 of the 24 months and gained 2.1% on average. It fell in October 2008 (-7.4%) and in December 2015, among others.
• Government bonds rose in 16 and gained 0.5% on average. Their gains were small relative to the equity losses they offset.
• Euro cash shows 9 losses, all tiny. They come from negative euro overnight rates between 2014 and 2022, not from price falls.

On average 9.0 of the 12 asset classes fell in each of these months; the range was 4 in April 2025 to all 12 in December 2015. Across all 87 falling months for MSCI World, at least 9 of the 12 classes fell together in 36.8% of them; in the 157 rising months, that happened only once.

The exception that changed after 2021

The averages above hide a break.

Source: Copio calculation; monthly returns 2006-2026 (CSV).

From 2006 to 2021, high-quality government bonds tended to rise when equities fell, as falling growth expectations pushed interest rates down. From 2022, inflation and rising rates were the common shock, and bonds fell in many of the same months as equities, including June (-1.6%), September (-3.4%) and December 2022 (-1.9%), three of the 24 worst equity months. Twenty months is a short sample, so the second column is a warning rather than a new long-run estimate. Our study of 2008 and 2022 explains the mechanism in detail.

What this means for a portfolio

A loss is most damaging when it forces a sale. If equities fall and the assets you planned to sell for spending, a deposit or a margin call fall in the same month, the loss becomes permanent. That is the practical reason to care about which holdings fall together, rather than about the number of holdings.

The data also limits what diversification can promise. Across these 20 years no asset class rose in every bad month, and the most reliable ones, gold and government bonds, delivered small average gains and long weak spells of their own. Gold took 24 years to regain its 1983 peak in euros. Holding them reduces the share of a portfolio exposed to the same shock; it does not prevent a broad loss.

Checks you can run on your own holdings

1. Sort your holdings by driver. Put every fund into one group: equities, high yield and other credit, government bonds, gold, cash, other. Use the index named on each factsheet, not the fund's name.
2. Add up the growth-sensitive share. Equities, high yield and commodities fell together in most of the worst months. Their combined weight is a better guide to your crash exposure than your equity weight alone.
3. Look up your assets' worst months. The 24 worst months (CSV) lists what each asset class returned in each one. Compare March 2020 and September 2022: they hurt different holdings.
4. Match your cash needs to what held up. Money you may need within a few years belongs in assets that did not fall with equities, and even government bonds failed that test in 2022.
5. Check the currency. All figures here are in euros. A fund priced in dollars, or hedged differently, can behave differently in the same month.

Limits of this analysis

• The window starts in January 2006, the first month every series is available, so it contains one financial crisis, one pandemic shock and one inflation shock.
• Index returns exclude fund costs and taxes. The global government-bond index is hedged into euros; an unhedged bond fund behaved differently after 2021.
• The expected values assume normally distributed returns, which understate how often large losses happen. Their purpose is to remove a known selection effect, not to describe the true distribution.
• Monthly data misses moves within a month, and a correlation says nothing about the size of the next loss.
• Past co-movement depends on the economic regime. It is not a forecast.

How can copio help you examine your own allocation?

Copio is in private beta across Europe, with early access by application. It looks through your funds into asset classes and places each on a return-correlation map as a main driver, a true diversifier, dead weight or a shock absorber, so you can see whether your holdings differ by driver or only by label.

A correlation matrix shows which holdings move together, and a crash test shows how a mix like yours would have fared in past crises. Copio provides educational analysis, not personalised advice; it does not place trades or choose an allocation for you. The free portfolio checker is a first step.

This article is published for information and educational purposes only. It does not constitute personalised investment advice. Past performance is not a reliable indicator of future performance.