How Diversification Actually Works

Why counting holdings misleads

Maya owns rather a lot of different things and is not sure whether that makes her diversified. It might not.

Suppose she owns twenty holdings that all rise and fall together. When one falls, they all fall, and the portfolio behaves exactly like a single holding twenty times over. The count says twenty; the behaviour says one.

Now suppose she owns four holdings that respond to different conditions. When one is having a bad year, the others need not be, and the portfolio's total is steadier than any of its parts. Four holdings, genuinely diversified.

What matters is not how many things you own but whether they go wrong together. The technical name for the degree to which they move in step is correlation, and it is the whole subject.

The part that seems too good to be true

Here is the property that makes diversification unusual, and it is worth stating carefully because it sounds like a free lunch.

Combine two holdings with similar expected returns that do not move in step, and the combination has roughly the average of their returns — but less variability than the average of their variabilities. The good years of one partly offset the bad years of the other, so the combined result is smoother than either.

That is genuinely something for nothing, and it is close to the only such thing available. You have not reduced the return you expect. You have reduced how much the path wanders around it.

The reason it works is not magic: it is that some of what makes any single holding move is specific to that holding, and specific things average out across many holdings. What does not average out is whatever they have in common.

What diversification cannot do

It cannot remove what everything shares. If a fall affects the whole market, owning more of the market does not help. There is a floor to how much diversification within one asset class can achieve, and that floor is whatever the holdings have in common. On the Indian record the market falls together often enough — why stock markets crash.

It fails when you most want it. This is the uncomfortable one. In a serious fall, holdings that normally behave independently often start moving together, because the thing driving the fall is affecting all of them and because people selling to raise cash sell whatever they can. So the protection thins out precisely in the episode it was bought for. That is not a reason to abandon it; it is a reason not to rely on it as the only defence.

It does not raise returns. Combining a good holding with a mediocre one gives you something in between. Diversification is a way of getting a given expected return with less turbulence, not a way of getting more return.

And it does not make a bad holding safe. Twenty poor holdings diversify away the differences between them and leave you owning the average of twenty poor things.

Where the diversification actually comes from

Not from owning more shares. The largest reduction in how much a portfolio moves comes from holding things that are not shares at all — money in deposits or short-term bonds behaves quite differently from equity, and that difference is much larger than the difference between two equity holdings.

This is why the split between shares and safer holdings does more for portfolio risk than any choice made inside the equity part, and why it is the decision to spend time on: choosing an asset allocation.

Within shares, the diversification available is bounded and the sources of it are few: companies of different sizes, different industries, and different countries — though the last brings a currency risk of its own, international equity and currency risk.

The two mistakes

Buying more funds and calling it diversification. Several funds holding the same companies is one portfolio with several sets of charges. This is the most common way of doing it wrong, because the fund count feels like evidence — and the arithmetic of overlapping holdings is in combining broad market indices.

Assuming things that were unrelated will stay unrelated. Correlations move, and they move most in exactly the conditions that matter. A pair of holdings that behaved independently for a decade can move together in a crisis, and a portfolio built on the assumption that they will not is more fragile than its history suggests.

How to tell whether you are diversified

Ignore the number of holdings and ask what they depend on. If most of your money depends on the same economy, the same currency and the same few large companies, you hold one thing in several wrappers, however many line items appear on the statement.

The practical version of that check — what your combined portfolio actually holds, and how much of it sits in the same names — is measuring concentration and overlap. How far to take it before the returns diminish is how much diversification is enough.

What to take away

Diversification works because holdings do not fail together, not because there are many of them. The benefit is a smoother path to the same expected return, which is real and unusual and worth having.

It is bounded by what your holdings have in common, it weakens in a crisis, and it is not a substitute for holding some money in things that are not shares. Get the last part right and the rest matters much less than it appears to.