Somewhere there is a person whose job is to order desks.
Not a metaphor. An actual purchasing officer at an actual education ministry, with a spreadsheet open, working out how many units of seating to buy for a building that won't need them for another six years. It's a boring job and it's normally a very predictable one, because the raw material announces itself well in advance. Babies get born, they turn six, and somebody in an office orders desks.
Then Singapore had 52,957 babies in 1988.
The year before it was 43,616. That's a jump of 21.4 percent in twelve months, in a wealthy, stable, obsessively measured city-state where the fertility rate had been sliding for a decade and went right back to sliding the moment this was over. Nothing had ended, nobody had discovered anything, and the policy was the same policy as the year before. Births fell to 47,669 the following year and kept falling.
That's the largest verified dragon-year spike on record. And every one of those nine thousand extra people was going to need a desk.
The crowded year, described from inside it
If you've ever been in a school year that was too big, you know it as a texture rather than as a statistic.
You get the second lunch sitting. The cafeteria was built for a normal year and yours isn't one. You get the portable classroom out on the field, the one with the metal steps and the air conditioner that sounds like it's being hurt. Your class is thirty-nine instead of thirty-one. Sports day has heats. There's a waiting list for the good elective and you're on it, and the person who tells you you're on it isn't unkind about it, just tired, because she's said that same sentence forty times this week and it's only Wednesday.
Nobody sits you down and explains that you're one of too many. You just notice, in a low-grade ambient way, that everything has a queue in front of it.
Which nobody ever names, because from the inside a crowded year doesn't feel crowded, it feels like the normal amount of world, and it isn't until years later, when someone four years younger describes a school day with actual room in it and a teacher who knew everyone's name by October, that you understand you spent an entire childhood in the overflow of a decision made by strangers before you were born.
Anyway.
And it doesn't stop when school does. The year that fills a cafeteria fills a university admissions round eight years later and a graduate hiring season four years after that, and it does all of this without anybody in it having done a single thing to earn the pressure. Vietnam's university entrance exam drew around 925,000 candidates in 2018, up roughly 60,000 on the year before, and a good share of that increase was the golden dragon cohort of 2000 arriving at the door together.
Nobody picks their year. You get handed one, along with everyone else who got handed the same one, and after that it's arithmetic.
The dragon, and the trap sitting in the dates
The Dragon is the fifth of the twelve animals of the Chinese zodiac, after the Rat, the Ox, the Tiger and the Rabbit, and it's the only one of the twelve that doesn't exist. Everything else on the list is something you could in principle be bitten by. The dragon is the exception, which is worth a second of your time, because a system otherwise assembled entirely out of livestock and vermin has one slot reserved for a thing nobody has ever had to feed.
It's also not the Western dragon, and this is where a lot of English-language writing on the subject goes wrong in its first paragraph. The long is a water creature and a rain-bringer, tied to rivers and storms and imperial authority, rather than a reptile that hoards gold in a cave until a knight turns up. The zodiac isn't a novelty item either. It's a live working calendar across China, Vietnam, Korea, Japan and the diaspora, used for reckoning years, for asking someone's age without asking their age, and for a genuine amount of household decision-making.
Recent dragon years: 1976, 1988, 2000, 2012, 2024. Next one is 2036. Want to check a year yourself? Dragon years divide by twelve with a remainder of eight.
Here's the part that catches people.
The year doesn't turn on January 1, and depending on which convention you're using, it doesn't turn on the same day as the other convention either. Popularly, the animal changes at Chinese New Year: February 17 in 1988, February 5 in 2000, January 23 in 2012, February 10 in 2024. In bazi, the four-pillars astrology used for actual chart work, the year pillar turns at lichun instead, the solar term that lands around February 3 to 5. In 2024 the gap between those two dates was six days wide, so anybody born between roughly February 4 and February 9 of that year is a Dragon under one system and a Rabbit under the other.
The 1988 gap is the brutal one. Chinese New Year fell on February 17 that year, which makes everybody born between January 1 and February 16 of 1988 a Rabbit, and there are a great many of them walking around with dragon merchandise they bought themselves. Seven weeks of people.
(This is the single most common error on the whole subject, and it's committed by roughly every gift shop on earth, plus a large fraction of the internet, plus the well-meaning relative who bought the enamel pin, so that a person born in early February of a dragon year can go their entire life being congratulated on a birthright that belongs to the animal before them, and will usually find this out from a stranger in a comment section, at thirty-four, having already had it engraved on something.)
Ugh.
The size of the bump, and the hole that follows it
The Singapore figure above is the headline, but it isn't alone, and the pattern across countries is consistent enough to be genuinely interesting.
Taiwan's Ministry of the Interior recorded about 424,000 to 425,000 births in 1976, up roughly 18.6 percent on 1975. In 1988 it was 342,031, up 8.9 percent. In 2000, 305,312, up 7.6 percent. In 2012, 229,481, up 16.7 percent. Hong Kong ran smaller and steadier, with 75,412 births in 1988 and 54,134 in 2000, up 7.8 and 5.6 percent respectively. Malaysia's Chinese-ethnic births went from 76,642 in 2011 to 85,419 in 2012, which is 11.5 percent.
So the effect is real and it repeats. It's also not enormous. Somewhere between five and twenty percent, depending on the country and the decade, which is a big enough number to buy desks over and a small enough one that you'd never spot it walking down a street.
The part that gets left out, and that changes what the whole thing means, is the year afterward.
Taiwan dropped in 1989. Hong Kong fell 7.7 percent in 1989 and 10.9 percent in 2001. Singapore's 1977 came in 10.3 percent below the dragon year, and Malaysia's Chinese births fell 18.6 percent in 2013. Every one of those holes is roughly the shape of the bump that preceded it, which is exactly what Yip, Lee and Cheung concluded from the Hong Kong data in 2002. The dragon year doesn't make babies. It moves them. Couples who were going to have a child sometime in the next couple of years have it in the dragon window instead, and cumulative fertility barely registers the difference.
Which means the whole phenomenon is a scheduling decision, taken at enormous scale, by millions of separate households that never coordinated with each other and produced a coordinated result anyway.
Worth saying plainly, because it's easy to be flippant about this from a distance: these are people making a real decision under real conditions. Timing a birth costs money and career momentum, and in a lot of cases it costs a conversation with a doctor. A study of Shenzhen birth certificates found roughly 6.7 percent of births shifted out of the week before a dragon year and into the week after, through both vaginal deliveries and elective caesareans, and found no measurable cost to the babies in birth weight, Apgar scores or neonatal mortality. Nobody here was being reckless. They were being hopeful, on a deadline, and negotiating it with a hospital scheduling desk.
What the crowd costs
Now the depressing literature. It came first, and I think it's been unfairly buried under the cheerful one.
Agarwal, Qian, Sing and Tan published a study in Population Studies in 2021 using Singapore administrative records, and it's the cleanest test of cohort size anybody has run on this. Chinese births rise 9.3 percent in dragon years in their data. The working paper puts the annual increase at 12.07 percent for Chinese births against 3.28 percent for non-Chinese ones, which is a lovely built-in control, since the calendar is identical for everyone and the belief isn't.
What happens to the dragon babies is that they earn 6.3 percent less. They're also 2.3 percent less likely to get into either of Singapore's two major national universities, and 3.4 percent less likely to end up in a professional occupation, with a slight bump in self-employment that in this context reads less like entrepreneurship and more like what you do when the graduate hiring rounds have already filled.
And here's the finding that rules out any explanation involving the dragon itself. Non-Chinese Singaporeans born in those same years take a hit too, around 3.8 percent on income, despite having no dragon belief and no dragon expectations attached to them by anyone. They were just in the building when it got crowded. So were the neighbouring cohorts who hit the labour market at the same moment. The mechanism is capacity. Too many people show up at the same set of institutions, and the institutions don't get bigger to meet them.
Sim found a similar negative effect on university attainment in Singapore in a 2015 paper, though that one carries a real methodological problem, because it treats everybody born in 1976 and 1977 as the dragon cohort when the 1976 dragon year actually ran from January 31, 1976 to February 17, 1977. That definition sweeps in Rabbits at one end and Snakes at the other. So, grain of salt.
Anyway. The crowd charges rent, and that part is well evidenced.
The study that found the opposite
Naci Mocan and Han Yu published "Can Superstition Create a Self-Fulfilling Prophecy? School Outcomes of Dragon Children in China" in the Journal of Human Capital in 2020, and it found dragon children doing better at more or less everything.
They're six percentage points more likely to hold a college education than children born in the neighbouring zodiac years, which is a 19 percent increase, and a regression discontinuity design run around the cutoff puts it at nine points. They score nine to ten and a half points higher on the gaokao, the national college entrance exam. That's only about 1.7 percent of the mean, and it matters enormously anyway, because the gaokao allocates university places strictly by rank inside one of the most competitive sorting systems ever built. In middle school they come out 0.04 standard deviations ahead in Chinese and 0.03 ahead in English, with maths positive and not statistically significant.
So the superstition works.
Except it doesn't, and the way they show this is the best thing in the paper.
They put parental expectations into the model. Specifically, what the parents said they expected their child to achieve educationally and professionally. The dragon effect on test scores then disappears. Not partly. Gone. Adding expectations cuts the coefficient roughly in half for Chinese and by about 80 percent for English, turns maths negative, and leaves none of the remainders distinguishable from zero.
Then they go looking for what the parents are actually doing, and they find it. Parents of dragon children are 1.4 percentage points more likely to contact a teacher five or more times a semester. They enrol them in kindergarten at higher rates and hand them more pocket money. They also assign them fewer household chores, which is a small thing standing in for a large one, because a chore is an hour that isn't homework.
The finding I haven't been able to stop thinking about is the height one. In their sample, girls are 6.4 centimetres shorter than boys on average, and a dragon girl closes 2.7 centimetres of that gap, with the effect larger in rural households (3.3 centimetres) than urban ones (1.9). Height at that age is substantially a nutrition story.
Somebody was putting more food on a plate.
The children themselves, though, aren't different in any way the researchers could find. Dragon children don't rate their own ability higher than other students and they're no more likely to say they expect a successful future. Their cognitive test scores match. Their parents look identical on education, occupation and income. Teachers can't see birthdates, and the gaokao is blind-graded, so nobody's being marked up for having been born in the right February.
The authors say it about as plainly as an economics abstract permits: neither the dragon children nor their families are inherently different from anybody else, and the belief in the prophecy, plus the investment that follows from the belief, becomes self-fulfilling.
Two literatures, one shelf
So you've got a paper saying dragon kids earn 6.3 percent less and a paper saying dragon kids are 19 percent more likely to hold a degree, and both of them are competently done.
They're also both right, which Chou demonstrated in Kyklos in 2022 using Taiwanese survey data, by finding a positive direct effect and a negative cohort effect running at the same time in the same population. One of those forces is your own parents and the other one is everybody else's. They point opposite ways. The net result depends on whether your country's schools and employers could absorb an extra nine percent of you.
Which is also why the same superstition produces a different answer in Singapore than it does in China. Not because the dragon behaves differently across a border, but because the number of university places does, and that number is set by a ministry with a building programme and a budget.
The five dragons, briefly, since somebody always asks
Each animal cycles through five elements on a sixty-year rotation. The element comes from the heavenly stem, not from the animal.
1928 and 1988 are Earth Dragons. 1940 and 2000 are Metal. 1952 and 2012 are Water. 1964 and 2024 are Wood, and 1976 and 2036 are Fire. The same element-animal pairing comes back every sixty years exactly, so 1904, 1964 and 2024 are all the identical stem-branch combination, jiachen.
One nuance that almost never survives translation into English horoscope writing: the branch chen, which is the dragon's, carries its own fixed element of Earth regardless of what the stem is doing. So 2024 is a Wood Dragon by stem sitting on Earth by branch. If you've ever wondered why two sources disagree about which element your dragon year is, that's usually why, and neither of them was lying to you.
The year it stopped working
2024 was supposed to be the big one. Wood Dragon, same combination as 1964, landing after decades of falling birth rates across the region. A lot of people expected a bump.
Taiwan recorded 134,856 births in 2024. That's 715 fewer than 2023, the lowest figure in the history of the ministry's records, and the ninth consecutive annual decline. Taiwan produced fewer dragon babies in 2024 than it did tiger babies in 2022, which had never once happened before, the Tiger being the year families have historically tried to steer around.
Fewer dragons than tigers. Nope, nobody predicted that.
Singapore's total fertility rate held flat at 0.97, making 2024 the first dragon year since 1964 to produce no fertility lift at all. Malaysia's Chinese births rose 0.2 percent. That isn't a rise. Hong Kong did climb about ten percent, to 36,700, but it also introduced a HK$20,000 cash bonus for newborns inside the same window, so take your pick of causes. Mainland China's births went up to 9.54 million from 9.02 million, and demographers there credited a post-COVID marriage rebound and expanded childbirth support at least as much as the dragon.
Then 2025 came in at 7.92 million, down 17 percent, the lowest since 1949.
So much for that.
So a superstition that could reliably move a national birth statistic for fifty years has mostly stopped moving it, inside a single generation, while the papers about its consequences were still going through peer review.
What's left of the prophecy
Strip out everything unverifiable. Here's what stays standing.
A belief with no mechanism attached to it moved millions of birth dates across five decades and several countries. It didn't create children, it rescheduled them, and the hole after each bump comes out almost exactly the size of the bump. Those rescheduled children then walked into schools and job markets holding too many of them and paid for it in admissions and in wages. And in the same population at the same time, a good number of those same children were being fed slightly better, pushed slightly harder, and represented to their teachers by a parent who phoned more often than other parents phoned, and they came out ahead on the one exam that decided everything.
The dragon did none of that. It hasn't got hands. Somebody ordered the extra desks six years early, and somebody else sat down at a kitchen table every night with a kid who'd been told since birth that they were the lucky one, and between those two people the whole prophecy got built by hand.
