Twice a year, in late March and late September, the Nikkei 225 and TOPIX open lower for a reason that has nothing to do with sentiment: thousands of Japanese companies go ex-dividend on the same morning. Each stock’s ex-date opening price is set by the market, but in theory it is lower by the dividend amount — and because the headline indexes make no adjustment for ordinary dividends, those lower prices flow straight into the index level. The aggregate drag is calculable in advance, and brokers publish estimates of it days ahead. If you know the three dates that govern the cycle — the record date, the last cum-rights day (kenri-tsuki saishubi), and the ex-dividend day (kenri-ochi) — you can separate a genuine sell-off from an accounting artifact in about thirty seconds.
The pattern was on full display in late September 2026: with thousands of interim record dates set for the September 30 fiscal half-year-end, the ex-dividend adjustments landed on the morning of September 29, and the Nikkei 225 finished that session at 65,481.27, with Tokyo Electron closing at ¥11,505. On September 30 — the final session of the quarter — the tone firmed, with 11 of 17 TOPIX-17 sector gauges advancing and Tokyo Electron gaining 3.17%. The mechanical drag into the ex-date is the calendar doing what it does every March and September; the firmer tape the day after was how this particular quarter happened to play out, not a sequel the calendar guarantees — whether the market recovers the drop depends on subsequent demand, as discussed below. The rest of this guide explains the machinery in evergreen terms.
Why Japan’s dividends cluster in late March and late September
Most listed Japanese companies run an April-to-March fiscal year. Dividend entitlement is determined by a record date (kenri kakutei-bi — the date on which you must appear in the shareholder register to receive the payment), and companies almost universally set it at the fiscal period end. That produces two enormous clusters:
- Late March — the fiscal year-end dividend (kimatsu haito), typically the larger of the two payments, plus companies that pay only once a year.
- Late September — the interim dividend (chukan haito) at the half-year mark.
Japan adds a second magnet that most foreign investors underestimate: shareholder benefit programs (kabunushi yutai — non-cash perks such as gift cards, food, or discounts granted to shareholders of record). These use the same record dates, and they attract retail buying that is insensitive to price in the days before the cutoff. The result is that a very large share of the market’s entire annual cash return concentrates on two mornings a year, rather than being spread across quarterly ex-dates as in the US.
The three dates: record date, kenri-tsuki, kenri-ochi
Japan settles equity trades on a T+2 basis (trade date plus two business days). To be a shareholder of record, your purchase must settle on or before the record date. Working backwards:
| Term | Japanese | Rule | Example: Sept 30, 2026 record date |
|---|---|---|---|
| Record date | kenri kakutei-bi | Set by the company, normally fiscal period end | Wednesday, September 30 |
| Last day with rights | kenri-tsuki saishubi | Two business days before the record date | Monday, September 28 |
| Ex-dividend day | kenri-ochi-bi | One business day before the record date | Tuesday, September 29 |
Buy on the last cum-rights day and you receive the dividend; buy on the ex-dividend day and you do not, which is precisely why the stock — and, in aggregate, the index — typically opens lower that morning. Count business days, not calendar days: Japanese public holidays shift these dates, and September in particular often contains multiple exchange holidays that compress the schedule. When the fiscal period end falls on a weekend or holiday, do not shift the record date yourself — the company’s stated record date stands, and the exchange entitlement schedule is adjusted so that eligible trades settle by that date. Rather than counting by hand around holidays, check the JPX ex-rights calendar for the official schedule.
Worked example: reading the September 2026 dates
Take the actual half-year end of September 30, 2026, a Wednesday, as the record date. Step back one business day: Tuesday, September 29 is the ex-dividend day. Step back one more: Monday, September 28 is the last day to buy with rights, because a trade struck that day settles on Wednesday the 30th under T+2. Now look at the tape: the Nikkei 225’s official close on that ex-dividend Tuesday was 65,481.27 (September 29, 2026, Nikkei Inc. figure). Part of whatever decline occurred that session was mechanical, not discretionary selling. For scale, if the aggregate dividend drop equalled roughly 0.4% of the index — a plausible order of magnitude for a September cluster, though the true figure varies each year and is published in advance by index providers and brokers — that alone would be roughly 260 points at a 65,000 level. The correct reading of an ex-day session is therefore: add the estimated dividend drop back to the headline change to find the market’s real verdict. Concretely, if the index falls 300 points on an ex-day carrying an estimated 260-point dividend drag, the underlying move is −300 + 260 = −40 points — essentially flat, not a sell-off. A headline decline can conceal a flat or even firm underlying market.
Why the index drop is mechanical — and calculable
Neither the Nikkei 225 nor TOPIX adds ordinary cash dividends back into the headline number. They are price indexes: each constituent’s ex-date price is market-determined, but it is theoretically lower by the dividend amount, and the index simply records that market price with no offsetting adjustment.
- Nikkei 225 is price-weighted with a divisor. The divisor absorbs corporate actions such as stock splits, but it makes no adjustment for ordinary dividends. On a concentrated ex-date, the sum of the constituents’ per-share dividends (adjusted for each stock’s price-adjustment factor) divided by the divisor is the theoretical mechanical point loss. High-priced constituents with meaningful payouts matter most.
- TOPIX is free-float market-cap weighted, and its headline version likewise ignores ordinary dividends. The drag equals the aggregate dividend value of constituents going ex, as a share of total index market cap.
- Both providers maintain total-return versions that reinvest dividends, and Nikkei Inc. publishes dividend-point data for the 225. Securities firms circulate estimated dividend-drop figures (yoso haito-ochi bun) ahead of each cluster — this is the number to have on your desk before the ex-day open.
One important consequence: the price index never earns those points back automatically. There is no rule that the market must recover the drop the next day. Over long horizons, comparing the headline Nikkei or TOPIX to a total-return benchmark such as the S&P 500 with dividends reinvested systematically understates Japan — always compare like with like.
The futures tell: seeing through the drop in real time
Index futures already discount known dividends in their fair value. Because a futures buyer receives no dividend, the expected drop is embedded in the basis (the spread between futures and spot) well before the ex-date. On the ex-morning, the spot index gaps down by roughly the dividend amount while futures do not — the basis simply converges. This gives you a practical, real-time check: if the spot index is down 0.5% on an ex-day but futures are roughly flat versus their previous close, the move is overwhelmingly mechanical. If futures are down as much as spot, something real is happening on top of the calendar.
The recurring flow pattern around the cluster
The calendar generates repeatable flows on both sides of the date:
- Pre-date accumulation. Retail investors buying for dividends and yutai perks, and dividend-oriented funds positioning for the payment, tend to support cash equities in the one to two weeks before the last cum-rights day. Stocks with popular perk programs often see the most price-insensitive demand.
- Ex-day mechanical gap. The index typically opens lower by roughly the aggregate drop; nothing needs to have changed fundamentally.
- Post-date disposal. Some of the pre-date buyers sell once entitlement is locked in, which can keep a mild headwind on cash equities for several sessions after kenri-ochi — a flow, not a signal.
- Passive reinvestment via futures. Index funds tracking the Nikkei and TOPIX will receive the cash dividends only weeks later, but to stay fully invested they buy index futures around the ex-date in size that market participants have historically estimated in the hundreds of billions of yen and, in the larger March cluster, on the order of a trillion yen. This reinvestment bid is one reason ex-days often trade firmer than the headline gap suggests.
- Fiscal period-end noise. The same late-March and late-September dates coincide with institutional book-closing, window-dressing, and rebalancing, so attribute moves to the dividend calendar only after netting out the published drop estimate.
Common mistakes to avoid
- Reading the ex-day gap as bearish sentiment. Add the published estimated dividend drop back to the headline change before judging the session. The dividend is not destroyed — it moves from price to shareholders’ pockets.
- Buying on the ex-dividend day expecting the payment. Entitlement requires owning through the close of the last cum-rights day, two business days before the record date.
- Counting calendar days instead of business days. Holiday-heavy months, September especially, shift the entitlement schedule. Work from the JPX ex-rights calendar rather than counting backwards by hand.
- Expecting an automatic bounce. The price index owes you nothing; recovery of the drop depends entirely on subsequent demand.
- Benchmarking the headline Nikkei against total-return indexes abroad. Use the total-return variants for any long-horizon comparison.
FAQ
If I buy a Japanese stock on the ex-dividend day, do I receive the dividend?
No. Under T+2 settlement you must buy no later than the last cum-rights day (kenri-tsuki saishubi), two business days before the record date, so that your trade settles by the record date. Buying on kenri-ochi means the seller keeps the dividend.
When is the dividend actually paid?
Not on the record date. Japanese year-end dividends are typically approved at the annual general meeting and paid roughly two to three months after the March record date; interim dividends for September record dates are generally paid around early winter. The cash lag is why index funds bridge the gap with futures.
Why don’t Nikkei 225 futures fall by the dividend amount on ex-day?
Because expected dividends are already discounted in futures fair value. The futures-spot basis absorbs the drop in advance and converges on the ex-date, which makes the futures market the cleanest real-time gauge of how much of an ex-day move is mechanical.
Is the March or September drop larger?
March is typically the bigger event: fiscal year-end dividends are usually the larger installment, and companies paying only annually cluster there as well. September’s interim cluster is meaningful but historically smaller.
Sources
- Japan Exchange Group — market statistics and trading calendar: www.jpx.co.jp/english
- Nikkei Indexes — Nikkei 225 methodology, divisor, and dividend point data: indexes.nikkei.co.jp/en
- Ministry of Finance Japan — securities flow statistics: www.mof.go.jp/english
Disclaimer: This is an information and analysis publication, not investment advice. See our Methodology for data sources, standards, and our corrections policy.