Diversification is often described as the only free lunch in investing.
Spread money across things that do not move together, and you reduce the damage any one of them can do without giving up much expected return. It is the premise behind every sensible portfolio, and it is genuinely sound.
There is also a problem with it that most consumer investing content never mentions, and it is worth understanding before you rely on it.
The Basic Idea
If you own one company and it fails, you lose everything you put in.
If you own five hundred, one failing costs you a fraction of a percent. The others continue.
The mechanism is that different holdings do not move in lockstep. When one falls, another may rise or hold steady, and the combination is steadier than any individual part.
The measure of how much things move together is called correlation. Lower correlation means better diversification.
Benjamin Graham connected this directly to margin of safety, noting the two ideas are correlative — each implies the other. A margin improves your odds on any single decision; diversification means no single decision can determine the outcome. We covered the first half in margin of safety explained.
The Problem
Correlations are not stable. They change, and they change in the least helpful direction.
Sébastien Page, who runs asset allocation at T. Rowe Price, has spent years documenting this. He gave a talk about it in 2008 titled “The Myth of Diversification,” and a paper of the same name won a major award in the field.
His finding, stated plainly: diversification does not work when you need it most.
In falling markets — particularly severe ones — correlations rise sharply. Holdings that behaved independently for years begin moving together. The protection you thought you had thins out at precisely the moment it is required.
The effect is well documented across individual stocks, country markets, industries, currencies, hedge funds and international bonds. It is not a quirk of one market or one period.
The Asymmetry Makes It Worse
Page and his colleagues found something else, and it is the part that stings.
Correlations do not just rise when markets fall. They also fall when markets rise.
Which means you get strong diversification during rallies, when you would rather everything moved up together, and weak diversification during crashes, when you needed things to move apart.
As Page puts it, nobody wants diversification on the upside. The pattern is exactly backwards from what an investor would choose.
The Statistician in the Oven
Page uses an illustration for why standard measures conceal this.
A statistician sits with his head in the oven and his feet in the freezer, and announces that on average he feels fine.
The average correlation between two assets, calculated across all conditions, is an average of extremes. It tells you very little about how they will behave in the conditions that actually matter.
A portfolio built on average correlations is built on a number that describes no real situation.
What This Looked Like in Practice
Two concrete examples.
2008. Research by Leibowitz and Bova examined a portfolio diversified across US stocks, US bonds, international stocks, emerging market stocks and real estate investment trusts — a textbook allocation.
During the crisis, its sensitivity to the stock market rose substantially, and it underperformed a plain portfolio of 60 percent US stocks and 40 percent US bonds by around nine percentage points.
The more diversified portfolio did worse. That is the failure of diversification in a single sentence.
2018. When the Federal Reserve raised rates and signaled more increases, all seventeen major asset classes tracked by Morgan Stanley finished the year down.
Every one. There was nowhere for asset class diversification to work.
Where It Fails Hardest
Worth knowing, because these are the things most often sold on their diversification benefits.
Alternatives and hedge funds. Page examined seven hedge fund styles, including supposedly market-neutral strategies. All showed considerably stronger correlation to stocks in falling markets than in rising ones. The strategies marketed hardest on their independence were not independent when it counted.
Private assets. Private equity and private real estate appear less correlated with public markets, and there is a reason that is misleading: they are valued infrequently. David Swensen, who ran Yale’s endowment and was a strong advocate of private markets, acknowledged the point directly — the apparent low risk is partly a statistical artifact. Two otherwise identical companies, one public and one private, will look different simply because one is priced daily and the other rarely.
Measured over longer periods, which reduces that smoothing effect, private assets diversify considerably less than reported figures suggest.
Even stocks and bonds. The most reliable diversifying pair can turn positive. Both are claims on future cash flows, so a shock to interest rates or inflation expectations can push them down together — which is what happened in 2018, and what makes the relationship more fragile when valuations are high in both.
What Page Is Not Saying
This is important, because it would be easy to draw the wrong conclusion.
He states explicitly that he is not arguing against diversification. His point is that standard measures overstate the protection it provides in stressed conditions, and that investors should calibrate expectations accordingly.
Diversification remains the right approach. It just is not body armor.
The Airplane
Page offers an analogy that resolves the practical problem neatly.
Pilots cannot predict when they will hit turbulence. Passengers can nonetheless be reasonably comfortable, because aircraft are built to withstand it.
You cannot forecast when correlations will spike — market falls are unexpected almost by definition. But you can build a financial position that survives them without needing to see them coming.
For an ordinary investor, that construction is not exotic. It is three things.
What Actually Protects You
Time. The single strongest protection available. If you do not have to sell during a downturn, correlations spiking is unpleasant rather than damaging. Someone with twenty years ahead of them experiences a crash as a period of buying at lower prices — the reasoning in Mr. Market.
Cash. An emergency fund is what stops a market fall becoming a forced sale. It is the reason the sequence puts it before investing, covered in how much you should have in an emergency fund. This is genuine crash protection, and it is uncorrelated with everything by construction.
Position sizing. Not owning so much of any one thing that its failure matters. This is diversification’s core function and it works in all conditions — a single company going to zero hurts you the same amount whether markets are calm or not.
Notice what is absent: nothing about clever asset mixes, alternatives, or products marketed on low correlation. Those are the things that disappoint under stress.
What This Means for a Normal Investor
Four practical conclusions.
Diversify — it still works. A broad index fund holding hundreds of companies protects you against any single one failing, permanently and reliably. That is the version of diversification that does not break, and it is the case made in index funds explained.
Do not expect it to prevent losses in a crash. Everything falling together is the normal pattern, not a malfunction. Expecting otherwise is how people conclude their strategy failed and abandon it at the bottom.
Be skeptical of products sold on diversification benefits. Complicated, expensive, or illiquid holdings marketed for their low correlation are exactly the category the research finds least reliable. If the pitch is that it moves differently from stocks, ask what happened in 2008.
Protect yourself with structure instead. Time horizon, emergency fund, and not overcommitting to anything. Unglamorous, and it works when correlations do not.
The Bottom Line
Diversification reliably protects you against one thing going wrong. That is genuinely valuable and worth having.
It protects you considerably less against everything going wrong at once, because in those conditions things stop moving independently. The research on this is extensive and consistent.
The response is not to abandon it — the person who documented the problem most thoroughly still recommends it. The response is to stop treating it as your only defense.
Your real protection against a bad market is not owning the right mix. It is not having to sell.
This article is general information, not personalized investment advice. All investing involves risk, including possible loss of principal.
Browse our other topics on the Explore BlurbMoney page.