Does the Share Price Really Drop by the Dividend?

The textbook says that on the morning a share goes ex-dividend, its price falls by the amount of the dividend. The money has left the company, so the shares are worth that much less.

We measured it across 31,577 ex-dates. The median fall was 0.89 of the dividend, which is close enough to the theory to call it confirmed. On 26.1% of those ex-dates the price went up instead.

Both of those statements are true, and holding them together is the whole point.

What the theory actually claims

A dividend is cash leaving the business. On the day the shares stop carrying the right to that cash, a buyer is getting slightly less than yesterday's buyer got, so they should pay slightly less.

This is why buying a share the day before its ex-date does not get you free money. You pay for the dividend in the price, receive it as cash, and end up where you started, minus whatever it costs you to trade.

That is the argument. It is correct, and it describes an average rather than a day.

On average, the theory holds up well

The measurement covers every regular dividend on our covered exchanges between January 2023 and June 2026, comparing the closing price on the last trading day before the ex-date against the close on the ex-date itself.

One detail in that sentence is doing a lot of work: closing price, not adjusted closing price. It is the single easiest way to get this measurement wrong, and it is worth a paragraph because anyone checking our figures will meet it immediately.

Most freely available price history is dividend-adjusted. Every historical price has been scaled down to account for the dividends paid since, which is exactly what you want for calculating a total return and exactly what you must not use here. An adjusted series has the ex-date drop removed from it by construction. Measure this with adjusted prices and you will find that shares fall by zero on their ex-dates, every time, in every market. That is not a finding. It is the adjustment you asked for, handed back to you.

The figures below use the traded closing price, corrected for share splits only.

The result is stable in a way that is quite reassuring. Whatever the size of the dividend, the median fall sits near nine tenths of it.

If the drop were an illusion, or an artefact of measurement, it would not sit in the same narrow band from the smallest dividends to the largest. It does.

On any given day, you cannot see it

Here is the part nobody mentions. The average is solid and the individual days are chaos.

Fewer than one ex-date in ten produces roughly the drop the theory predicts. One in four goes the wrong way entirely. The middle half of all outcomes ranges from the price rising slightly to it falling by nearly twice the dividend.

Nothing is wrong with the theory. The adjustment is real and it happens every time. It is simply much smaller than the ordinary daily movement of a share, so it is buried.

Which is why the size of the dividend decides the visibility

The clearest evidence for that explanation is what happens when the dividend gets big enough to see.

A dividend worth less than half a percent of the price is smaller than a quiet day's trading, so the direction of the day is close to a coin flip. A dividend worth more than 4% is a step nothing else on a normal day matches, and the price falls almost every time.

The size of the drop does not change. Only whether you can find it under the noise.

The same effect explains the differences between markets

This is also why the numbers look different by exchange, and why that difference is less interesting than it first appears.

The median fall barely moves: every exchange sits between 0.80 and 1.01. The share of ex-dates where the price fell swings much more widely, from 68.8% in New York to 90.3% in Brussels.

That is not a statement about market quality. It follows from what these markets pay. A continental European company paying its whole year in one instalment is handing over a large enough sum to clear the noise, so the fall is visible almost every time. An American company paying a quarter of a modest yield is not. The underlying adjustment is the same in both places.

So why does dividend capture keep getting recommended?

The strategy is to buy just before the ex-date, collect the dividend, and sell. Because the price falls by slightly less than the dividend on average, it looks like free money sitting in that gap.

Two things close it. The first is costs: two trades and a spread, on a gap worth a tenth of one dividend. The second is much larger, and it is the table above. The outcome on any single attempt is dominated by where the share happened to move that day, and the middle half of outcomes spans a range many times the size of the edge being chased. You would need to run it hundreds of times for the average to show up, and pay costs on every one.

An edge you cannot detect in fewer than several hundred attempts, on a gap of a few percent of a dividend, is not an edge. It is a hobby.

There is a fair objection worth conceding. The median fall really is below 1.00, at 0.89, and a shortfall of a tenth of a dividend is not nothing across a large enough sample. We cannot tell you what produces it. Candidate explanations range from the relative tax treatment of dividends and capital gains, which differs by country and by holder, to plain trading friction, and our measurement separates none of them. What we can say is that whatever produces the gap, it is far too small and far too noisy to be harvested by an individual paying retail dealing costs.

Does it matter when you buy, then?

Not for this reason, no.

If you are buying a company you intend to hold, buying the day before the ex-date and buying the day after leave you in materially the same position. One gets you a slightly higher price and a dividend; the other a slightly lower price and no dividend. On average those net out to within a tenth of a dividend of each other, which on a 2% yielder is around 0.05% of your position.

That is smaller than the spread you paid to trade. It is not worth planning around, and it is certainly not worth delaying a purchase you want to make anyway. The one practical consequence is administrative rather than financial: buying just before an ex-date gives you a taxable cash payment almost immediately, which some people prefer to avoid and others actively want. That is a preference, not an edge.

What the ex-date is actually for

Entitlement. That is the whole job.

Owning the shares before the ex-date means the payment is yours, whenever it arrives, and it does not have to arrive soon. For a European company the wait between going ex and being paid runs from two days to a month depending on where it is listed, and the month the cash lands in is what an income plan actually runs on.

If you are new to this, the beginner's guide to dividend investing in Europe covers how the dates fit together. And if you are choosing what to hold rather than when to buy it, the screener sorts on the things that decide whether the dividend survives, which matters considerably more than which Tuesday it goes ex.

One market we left out

Toronto is excluded from every figure above. Its listed property trusts produce a median price change of exactly zero on their ex-dates, which pulls the whole exchange to 0.60 and is not a believable market result. We read it as a problem with our own price data for those units rather than a fact about Canadian shares, and would rather leave the exchange out than publish a number we cannot stand behind. Toronto excluding property trusts sits at 0.77, closer to everywhere else but still low enough that we are not confident in it.