How a 30-year dividend streak is counted: the definitions behind one number

Many financial metrics have tricky formulas. ROE changes with what goes in the denominator, and interest coverage changes with which profit figure is used. Next to them, consecutive dividend years look easy at a glance: count back through the years a dividend was paid and stop where the chain breaks.

Questions come up as soon as the counting starts. It is August now; does this year count? A company that pays once a year has probably not paid this year's dividend yet, so counting this year would give it 0 years no matter how long it has kept paying. Including this year unconditionally, on the other hand, gives an extra year to a company that has already stopped paying.

Settling one standard brings the next question. Is a dividend year the year the money arrived, the year the right to the dividend was fixed, or the fiscal year whose results the dividend was based on? And for a company that usually pays nothing but now and then hands out a large sum, does that year count as one year of the streak?

If only the number is written down without saying how these questions were answered, a reader has no way to check it. This article sets out how Jupick answered them and where the limits are.

Jupick counts only through completed years. As of August 2026, the last completed year is 2025, and counting runs backward from there. Whether a dividend was paid in 2026 does not enter the count.

The reason is simple: the year in progress has no result yet. A company that pays every quarter already has a record for this year by March, but a company that pays once a year pays in the second half, so its record stays empty through the first half. Counting this year would mean measuring two companies on the same day and penalizing only one of them for a year that has not ended.

Institutions that build dividend indexes overseas use the same approach. A typical example is Nasdaq, which looks only at records for completed years when it builds its Dividend Achievers indexes. Leaving out the year in progress makes the number up to one year more conservative, but in return the answer is the same whenever it is looked up. For this metric, a value that does not differ between March and September matters more.

So a company shown on Jupick with a 30-year streak has kept paying for 30 years through last year; it does not mean it has paid this year. Whether it has paid this year is more accurately checked in the dividend schedule.

They appear under the same item, but the years for Korean stocks and overseas stocks come from different records. It is better to say so up front than to hide it.

For Korean stocks, Jupick reads the annual reports filed on DART, Korea's electronic disclosure system. Each report states the cash dividend per share for that fiscal year, and Jupick goes back from the latest report, counting the years in which that value is above 0. The Korean standard is therefore the fiscal year. Because these are figures the company disclosed itself, the basis is clear, but a report takes several months to come out.

For overseas stocks, Jupick reads the record of ex-dividend dates. The ex-dividend date is the reference date for the dividend right: shares have to be held before it for the holder to be entitled to the dividend. Because what is counted is the year in which this date falls, the overseas standard is the calendar year from January to December. There is no wait for filings, so the latest state shows up right away, but it can fall out of step with the company's fiscal cycle.

The two methods usually give similar answers, but at the edges they can differ by about a year. That happens with companies whose fiscal year does not end in December, companies that changed their fiscal year, and companies whose dividend was paid in the following year. Setting the years of a Korean stock against those of an overseas stock one year at a time to rank them is therefore not a sound use of the number. A gap between 15 and 16 years can be a difference in the ruler, not in the company.

International statistical bodies keep a separate mark for such cases. Eurostat, the statistical office of the European Union, attaches a “definition differs” flag to a value when the same item is defined differently from country to country. The method neither deletes the number nor pretends it is the same thing; it simply records that it differs. Writing this into the cautions on the glossary page follows the same idea.

Not all the money a company pays out has the same character. Regular dividends are paid every quarter or every year under a set policy, and special dividends are paid once, when the company has disposed of assets or built up a large pile of cash. By their nature, special dividends carry no assurance of continuing into the next year.

Since what a dividend streak tries to measure is consistency, strictly speaking it is more accurate to leave special dividends out of the count. In fact, the institutions that build dividend achiever indexes do exclude them.

The problem is that the ex-dividend records for overseas stocks carry no mark separating regular from special dividends. Freely available data has only the date and the amount. An unusually large amount can be guessed to be a special dividend, but once records start being corrected on such guesses, they stop being data and become estimates. So Jupick counts them as they are, without separating them.

There is one visible sign of this limit: a stock with a dividend yield of 0% that still shows consecutive dividend years. This fits a company that pays no regular dividend now but has paid only special dividends in recent years. For such a stock, it is better not to read the years as evidence of consistency and to check the whole dividend section once more.

A dividend streak starts over when it breaks. A company that kept paying for 30 years and then skips one year shows 1 year the following year, not 29. Taking a long time to build and only one miss to lose is the nature of this metric.

So a value of 0 years gathers very different situations in one place. A company that has never paid a dividend shows 0 years, a company that paid for a long time and recently stopped shows 0 years, and a company that has just started paying and has not yet completed a full year also shows 0 years. The three companies are in completely different positions, yet the number is the same.

The 0 years of a growth company that pays nothing so it can reinvest its profit cannot be viewed the same way as the 0 years of a company that dropped its dividend after its results worsened. This is why, when the value is 0, the number alone is not enough: earnings, the payout ratio and whether there was a past dividend record need to be checked as well.

A long streak is not complete information on its own either. Never skipping and steadily raising are different stories, and the years keep piling up even while the amount shrinks a little every year. The years tell only that a dividend policy has been maintained, not whether the dividend was generous.

Anyone who handles data runs into cases that cannot be counted. A data source stops responding for a while, or a delisting or a change of ticker code makes the stock impossible to look up. What to write down at that point is actually the part that calls for the most care.

The easiest choice is to write 0, and it is also the most dangerous. 0 is a real value for this metric. If a 0 meaning “unknown” and a 0 meaning “never paid a dividend” look exactly the same on screen, neither the scoring side nor the reader can tell them apart. Not knowing gets quietly translated into a bad score.

Jupick ran into this problem too. Judging that there was no source for the consecutive dividend years of overseas stocks, it filled them with 0 for a long time, and that 0 received the lowest score on the scoring table. Because the dividend area is the average of three metrics, every overseas stock was losing the same amount of dividend score. Missing data had turned into wrong data.

Now Jupick separates the case where it cannot count from the case where it counted and got 0. A company that was counted and came out at 0 is recorded as 0, which is a measured fact. A stock whose data could not be obtained is scored with a neutral fill-in value that neither helps nor hurts the score, and the record carries a mark that the value is not a measured value. The purpose is to exclude such values from the sample later when producing statistics such as metric distributions.

The mark also records the reason. An item the company simply does not have and data that was not received this time call for opposite actions from the reader. The first will still be missing if checked again tomorrow; the second is a value the next update will fill in.

Consecutive dividend years look simple, but in practice the value is built from a stack of choices: which year to count from, what counts as a dividend, and what to write when there is no data. When different sites show different numbers for the same company, it is usually because these choices differ.

So a number that differs from another site does not mean one side is wrong. What matters is whether the site says what it counted and how. A number with no written definition, large or small, cannot be checked.

Jupick's standards, gathered in one place, are these: it counts only through completed years; it uses the fiscal year of the annual report for Korean stocks and the calendar year of the ex-dividend date for overseas stocks; it cannot separate special dividends for overseas stocks; and when it cannot count, it records a neutral fill-in value and a mark instead of 0.

How far back the count goes also differs between Korean and overseas stocks. For overseas stocks it counts up to 30 years, as far as the ex-dividend record goes, while for Korean stocks it looks back through the annual reports one year at a time, so the number of lookups has a cap. This is another reason the years of Korean and overseas stocks are not to be compared by size.

That cap was 10 years for a long time, and the number hid a problem. The scoring curve Jupick uses for consecutive dividend years gives full marks at 15 years. If lookups stop at 10 years, no Korean stock can get full marks, however long it has paid dividends. The reason for missing full marks becomes how far Jupick looked, not the company's dividend history. In fact, 92 (46%) of the major Korean stocks were bunched at exactly 10 years. When nearly half of the values sit at one point and that point happens to be exactly where the counting stopped, it is not a distribution but a cut mark.

So the lookup cap was aligned with the full-marks point of the scoring curve, at 15 years. Widening a cap still leaves a cap, though. A value that reaches 15 years is still a lower bound, not an exact count, so such values now carry “or more” and are shown as “15 years or more.” The same goes for the 30 years overseas. A truncated value is not erased into a blank because it is not a missing value. Not knowing because there is no data differs from having clearly counted up to 15 years and not knowing what lies beyond. The first says nothing, while the second can say at least this much.

One more point: Korean financial data is not downloaded again for all stocks at once but refreshed in batches over several days. So when a counting standard like this changes, it takes several days for the result to show across the whole screen. Right after the change, values counted under the old standard may still be mixed in.

Consecutive dividend years show only one property of a dividend: its consistency. The dividend area reads properly only together with dividend yield, the payout ratio and the earnings behind the dividend. In the grade, dividends carry a weight of only 25%.

This content is for information only and is not investment advice. Jupick does not suggest trading in any stock and does not forecast future dividends or share prices, and a long or short streak of consecutive dividend years is not a suggestion of any kind. Investment decisions and responsibility for them rest with the user.