Guides · Markets & investing
Diversification and What It Can't Do
Diversification is called the only free lunch in finance. It's real, it's powerful, and in 2022 it very publicly did not save anyone. Both facts are true, and the reason is worth understanding.
Two risks, and only one of them goes away
Idiosyncratic risk is specific to one company: an accounting fraud, a failed drug trial, a warehouse fire, a CEO who turns out to be a disaster. Own thirty companies and any one of these damages a thirtieth of your portfolio instead of all of it.
Systematic risk is the market itself — recessions, rate shocks, pandemics. It affects everything at once, and no amount of spreading escapes it, because there is nothing within the market to spread into that isn't also exposed.
That's the precise claim diversification makes, and it's narrower than the slogan suggests. Free lunch, yes: you reduce idiosyncratic risk without giving up expected return, which is genuinely rare. But it was never a promise about market-wide declines, and the people disappointed by it in a crash were disappointed by a promise nobody competent made.
Company-specific risk goes away. Market risk does not, and cannot.
Count is not the measure — correlation is
The benefit curve flattens fast. Going from one stock to twenty removes a great deal of company-specific risk; going from thirty to five hundred adds relatively little more. Treat the exact number as contested rather than settled — Evans and Archer put it near 20 in 1968, Statman argued 30–40, and later work argued it rose as individual stock volatility increased.
But the count is the wrong thing to watch. What makes diversification work is low correlation — owning things that don't move together. Twelve stocks responding to the same interest rates, the same regulation and the same narrative is closer to one bet held twelve times.
This is also why holding your employer's stock deserves special caution. Your salary and that position share a single point of failure: the year the company struggles is the year your job is least secure and the shares are worth least, arriving together.
Why 2022 didn't disprove any of this
Stocks and bonds usually diverge, and historically bonds cushioned equity declines. In 2022 both fell — the Bloomberg US Aggregate lost about 13%, its worst year on record — and a lot of people concluded diversification was a myth.
The cause was specific rather than mysterious. The Fed hiked rapidly to fight inflation, and rising rates mechanically push bond prices down while also compressing stock valuations. One force hit both assets through different channels, so the usual offsetting relationship didn't apply.
The honest framing is that diversification is probabilistic, not a guarantee, and correlations rise in crises — precisely when you most wanted them low. That's a known limitation of a tool that still works, not a broken concept. Anyone selling it as a promise you'll never have a bad year is misleading you.
Diversification removes company-specific risk, not market risk. Correlations rise in crises, which is exactly when you wanted them low.