If almost everyone knows that Chinese firms cut anyone older than about 35, and after that all you can get is outsourced or dispatched work, and by 40 you are delivering meals or driving for a ride-hailing app, and past 50 you are cleaning offices or sitting a guard booth — then how many people would you guess rush out to buy a house and a car, get married, have children? One side talks about building a fertility-friendly society while doing the exact opposite. Some companies are plainly profitable and still run on outsourced and dispatched labour to “cut costs and raise efficiency”. Pitiful, sad, hateful. Given that expectation, I will just save and not spend, and at 35 retreat to a second- or third-tier city and live small there. Why would I spend my body grinding for this little money, playing your game? I am already on the bus. I have no choice now. But I will definitely tell my child how this thing eats people.
I wrote it angry. Afterwards I sat looking at those three numbers: thirty-five, forty, fifty. I do not need to check them, ask about them, or wait for them to happen. I know them by heart.
The 35 course I have actually eaten. The delivery years at 40 and the cleaning years at 50 I cannot speak to — they have not reached me, and nobody can give me a real figure. The one concrete story that keeps getting told is a job fair in Dalian in March 2025: a 38-year-old Java developer sent 119 applications in half a year and got 3 interviews.
One true story does not prove the whole road. But my decisions do not rest on that sentence being accurate. They rest on it being believable. Those are two different things.
Saving instead of spending — I did the arithmetic
I do not change cars, I take on no extra debt, I do not renovate, and I read the price of my kid’s after-school class three times before signing. Ask me why and I cannot explain it. I am simply afraid. I was not born frugal. I am saving for the future version of me who has no salary.
That version of me will never appear on a payslip. He can only live off the money set aside now.
So the column labelled “weak consumer demand” is filled with people like me doing sums late at night. It is not mood. It is arithmetic.
What the other side reads is never my feelings, it is my behaviour. I do not sign the purchase contract, I do not marry, I do not have children — every one of those motions is countable. Over there they call it a missed target. From where I sit it is “next month’s pay is not guaranteed”. Same event, two vocabularies.
On having children, I refuse the “young people are selfish” explanation
China’s National Bureau of Statistics, January 2026 release: 7.92 million births in 2025, 11.31 million deaths, total population down 3.39 million.
The UN’s World Population Prospects 2024 medium scenario projects China losing 155.8 million people over the next 25 years — the largest absolute decline in the world; second-place Japan loses only 18 million. That is a forecast, not a fact. I use it for magnitude, nothing more.
What actually convinced me is one distinction in an IMF piece. This round of decline is not like the Black Death or a war — not a mortality shock. Last time humans had fewer children it was because the children did not survive. This time the grown-ups decided it.
The same IMF authors admit that “fix employment and housing and the birth rate comes back” does not hold. Two of their reasons are very concrete. A child is itself an economic pressure: one more is not one more worker, it is eighteen years of spending you cannot cancel. And the extra labour power takes twenty years to arrive.
Lay that time gap out flat: if this generation does not reproduce, the bill surfaces in twenty years; if this generation does not reproduce, the cost is settled this month.
I am not saying that sums correctly. I am saying that a man with a thirty-year mortgage sitting where I sit will come out with exactly that answer.
The civil-service side changed the 35 line. I waited a long time for that, and still — I am on the wrong side of the door
The facts first. The 2026 national exam raised its cap from “under 35” to 38, with fresh master’s and doctoral graduates allowed to 43. Beijing, Shanghai, Tianjin, Jiangsu, Zhejiang, Sichuan, Guangdong and Henan followed; the Beijing notice literally reads “aged 18 to under 38”; more than twenty provinces sat a joint exam date and generally moved to 38 as well.
When I read the notice I laughed through my nose. You people held that 35 line for thirty years, issued document after document — did any of it reach the firms? But this time it is real, black and white, and I am not going to perform cynicism about it. This step, I take it.
Then I thought of my colleagues. This 38 door opened for other people: still queueing at the gate of the establishment, fresh master’s and PhDs. I was already cut once at 35; my résumé does not get past the screening system any more — the door opened, and it is not my cell. I am standing outside it. I cannot get in, and I have no bitter joke left to make.
Whether private employers moved with it, I have no data. The most recent quantitative measurement uses job-posting data from 2021; the perception surveys stop at 2023; nobody has re-measured whether the gate at internet companies loosened across 2024–2026.
And when firms screen you out they never say “age”. They say the experience does not match, the team needs to be younger, this role needs someone who can grind. Every one of those phrases is legal.
A paper says the age gate is a miscalculation, not cleverness
Caijing Studies, issue 1 of 2026, a Tsinghua PBCSF team. They built a city-level “age-discrimination-at-35 index” out of the difference in postable vacancies between age 34 and age 35, plus regex hits on text that says “no older than 35”.
The regressions: where discrimination is 1% higher, firms’ labour productivity averages 0.082% lower, total factor productivity 0.114%–0.119% lower, and cash pay per employee 0.134%–0.233% lower. The authors call it a loop — low human capital, low efficiency, lower human capital.
I felt no vindication reading this. I always assumed the age gate was a calculated move, shrewd. This says it was not calculated, or was calculated wrong: the people it cuts are precisely where its own productivity comes from, and the wage bill it saves gets paid back later.
That also explains my line about profitable companies still outsourcing to “cut cost and raise efficiency”. It is not greed. Greed at least knows what it is trading for. The machine can only read this quarter’s report. It cannot read me.
Where that 42.4% came from
The Journal of World Economics, 2023, vol. 46 no. 11, 鄢伟波. National business-registration data plus the 2008–2016 tax survey, chopping labour dispatch from 1994 to 2021 into five phases: policy carry-over, acceleration, explosion, regulation and diffusion, contraction.
The hardest finding: the 2008 Labour Contract Law increased registered labour-dispatch firms by 42.4%.
I read that three times. Just as the law tightened regular employment, dispatch grew by more than forty percent. On one side the statute said give them open-ended contracts, give them insurance, give them severance. The market’s answer was neither compliance nor violation — it was to change the entry labelled “employer”.
The 2013 amendment squeezed it for a while and pushed costs up. Then from 2015 it went back to rising. The author’s own conclusion: regulation did not reverse the expansion.
I will state the limits: that paper’s tax data ends in 2016, its phase narrative in 2021. So it does not prove dispatch is still growing today. It proves dispatch was squeezed once and grew back. Legislation can forbid using people this way; it cannot forbid using them this way under a different name.
Another set of numbers, from the Guanghua School report on the 2025 labour market: flexible workers show “three lows and three highs” — low insurance enrolment, low contract signing, low hourly pay; high intensity, high occupational risk, high probability of losing a claim. It puts their social-insurance coverage at roughly 20%, against 60–70% for conventional employees. That 20% is probably the employee-scheme definition; add resident insurance and it goes up. I will not die on that hill. The direction is cold either way: these people are not unemployed. They are working on the unprotected side.
Everybody feels it. Almost nobody sues.
The Chinese Social Survey has the share calling age discrimination “serious or very serious” at 20.6% in 2015, 37.5% in 2019, back to 33.3% in 2021, and it rises with age in an inverted U, sharpening from about 42. A 2023 Q1 Zhaopin survey: 85% believe a “35 threshold” exists, 46.8% say it is hard to find work past 35, and 60.2% rank age discrimination the top employment-discrimination problem needing fix — ahead of education at 42.2%, gender at 24.6%, household registration at 21.6%.
I put those side by side for one reason: the feeling is real, and the feeling is not rare.
On the other side: a Workers’ Daily reporter searched the court-judgment database and found that as of March 2025, labour disputes using the keyword “age discrimination” in the past ten years total 41 cases.
I have to attach two caveats or I will not trust the number myself. It is a keyword floor, not a census — it counts only cases where someone wrote “age discrimination” into the judgment. And since 2021 far fewer judgments are published at all, so today’s denominator is much smaller than ten years ago.
Even with both caveats it is still the same fact: a gate most people acknowledge, that almost never reaches a courtroom.
Why not? Because it does not travel that road. It sits on the job listing and you are filtered at the moment you apply — you do not even receive a rejection. It sits in the phone call before the offer. By the time you can actually sue, their stated reason is experience mismatch or team-refresh. What you would have to prove is “because I am thirty-five”, and that wall is higher than the gate.
Japan wrote the age gate into law as illegal. Then what?
This is the finding that kept me in the chair.
In Japan, putting an age limit on a job ad is illegal — not a guideline, a statute: the Work Force Availability Development Act, the old Employment Security Act, a legal duty on employers since the 2007 amendment. And the health-ministry’s official interpretation goes further than the text: writing “age not considered” on the request form while actually refusing applicants by age, or deciding pass or fail by age at document screening or interview, is equally illegal. Even a condition that converts people above a certain age to part-time regardless of their own wishes is prohibited.
The one lawful “35” window is Exception 3(i), and its conditions are very narrow: the contract must be open-ended, prior job experience must not be asked about, and non-graduates must get the same training and development as fresh graduates.
Japan does not say you may not compete on age. It says: to screen legally by age, you must pay for it with an open-ended, long-term regular post. Want to filter? Trade long-term employment for it.
Did it work? Partly. Here is the curve I copied: the age ceiling above which more than 40% of job-changers get a raise moved from the early 30s range in 2012 (under 30% past 40) to the late 40s range in 2024; the NISS Research Institute’s own conclusion is that, judged by wage movement, the line that once sat at 35 has been lifted to 50. In 2013, of people who changed jobs, seven in ten were under 45 (7:3 men, 8:2 women); by 2025 both sexes were 6:4. Of people placed through public employment offices in 2024, 73.1% were already employed — the highest since 2012. Of new jobseekers in 2025, those over 65 rose from 7.9% in 2015 to 19.1%.
So “35 is the end” was dismantled halfway in Japan. The gate was made illegal, and labour shortage pushed it back more than a decade.
And then comes the ugly part. Japan did not turn the story around.
Its non-regular employment is frozen at over a third: in 2025, 21.28 million people, 36.5% — and regular and non-regular grew at the same time (regular employment rose for eleven straight years from 2015). The money gap widens with age: men aged 45–49 in regular posts earn roughly 6.66 million yen a year, non-regular about 3 million — more than double. That status is worth more than a degree: a high-school graduate in a regular post generally out-earns a university graduate in a non-regular one.
The health ministry wrote the causal chain itself. The wage curve for regular employees flattens across the thirties and forties, and the ministry links that — income hardly rising exactly when spending peaks — to suppressed consumption and to hindered family formation: marriage and childbirth.
Half of the question I have been circling gets answered by an official document here. It is not “a generation turned lazy”. It is that the decade you most need money is the decade your curve will not move.
One more number I have to write down even though it cuts against my story: Japan’s “involuntary non-regular” (non-regular because no regular post was found) fell from 3.42 million and 19.2% in 2013 to 1.73 million and 8.4% in 2025. The coerced share is shrinking; the chosen share is rising. So “anyone over 35 gets pushed to the side” does not explain the whole expansion of non-regular work — including the table I wrote for myself.
The endpoint moved somewhere else in South Korea
Korea runs its non-regular supplement once a year, on an August reference date, so everything below is that basis.
August 2024: of 22.143 million wage employees, regular posts fell 147,000 to 13.685 million and non-regular rose 337,000 to 8.459 million, pushing the non-regular share to 38.2%. One component kept me staring: part-time (시간제) reached 4.256 million, up 383,000 — for the first time since records began in 2003, over half of all non-regular work (50.3%). Of the growth in temporary positions, 95.8% came from part-time; atypical jobs actually fell by 54,000.
Non-regularisation is sliding into something smaller. This is not staff turning into contractors. It is contractors turning into people paid by the hour.
The pay-gap line has an honest reversal in it, and I will not quote only the frightening half. In August 2024 the monthly gap between non-regular and regular reached a record 1.748 million won (3.796 million against 2.048 million; non-regular broke 2 million for the first time since 2003). But Statistics Korea states the widening is mainly a composition effect of part-time work — exclude part-time and the gap is 839,000 won, the smallest since 2018.
So that “record” cannot stand alone. It grew because the survey absorbed hundreds of thousands of low-paid hourly posts, not because the same jobs were squeezed harder.
Where is Korea’s 35? It is not at 35. In August 2024, of non-regular workers, 33.2% were aged 60 or over, 19.6% in their fifties, 17.3% in their twenties; women hit a record 57.3%. Statistics Korea attributes growth in manufacturing non-regular work to men over 60 being re-hired after retirement and thereby non-regularised.
The line actually sits there: not thrown out at thirty-five, but sliding back in after the retirement age, in non-regular status. That is far more concrete than anything I could make up, because it has an official basis.
On whether the old take the youngs’ places: Korea legislated a statutory retirement age of 60 in 2013, phased in from 2016, and three papers contradict each other. 한요셉 (2019, KDI) found a negative effect on youth employment — in the private sector, roughly one extra elderly post costs 0.2 youth posts. 이환웅 (2023, KIPF) found overall employment positive, with any crowding-out landing on ages 45–54 rather than the young. 강신혁 (2025, KLI) found elderly workers and young ones are not close substitutes.
What those three gave me was not a conclusion but the absence of one — all of them run into the same background: Korea’s population aged 19–34 shrank by 2.67 million between 2000 and 2020 and is projected to fall to 5.21 million by 2050, half the 2020 level, down to 11% of the population. The young are disappearing as a category, so “displacement” is being cancelled by demography.
That rebuts me too. I assumed that clearing out the old makes room for the young. The Korean and Japanese numbers say the cleared-out slide into non-regular work, and the young people’s openings did not grow either.
Set the three countries side by side and the shape is the same
Japan deleted the “age” column, so the gate moved to the regular/non-regular column. Korea’s gate moved to the retirement-age column. China’s gate still says thirty-five, but the column behind it says outsourcing.
I do not want to use Japan and Korea to comfort myself with “endure it and it passes”. Japan endured it — the gate is illegal, the line is at 50 — and suppressed consumption and delayed family formation are still written up as problems in a ministry document. The gate changed address; the machinery did not.
So moving thirty-five to forty will not save me, and replacing this generation will not either. What has to change is that other column: establishment, the housing-fund contribution base, whether annual leave exists, whose liability a work injury is, whether a non-renewal needs a reason.
I am already on the bus
After all those numbers I have to say where I stand, otherwise it looks like I am doing other people’s accounting for them.
I am already on the bus. Thirty-year mortgage signed, child in school, and my wife’s former employer owed her a full year of salary — that lawsuit is still running. “Refusing to cooperate” is a choice for someone not yet on board. For someone like me it is only an explanation.
I do not save because I like saving. I arrange every expense as if this job could end any month.
I will tell my child how this thing works. The reason is plain: the poison in an expectation like this is not whether it is true. It is that it gets handed down only in halves. The generation before taught only how to get in — which school, which company, which certificate. Nobody taught how you get priced once you are inside. So everyone bets from half a map, and when they hit a wall they blame their own effort.
In the end I still have no conclusion
I cannot prove that “35 is where the career ends”, and I cannot afford to disprove it. I only know what it collects: big purchases, a marriage certificate, a child, and any schedule a person dares to lay out twenty years ahead.
People ask what you do once you see it clearly. I go to work every day anyway. I make the mortgage payment on the tenth anyway.
I cannot change that 35. I can only rearrange the little budget I still hold according to it. There is nothing inspiring about that, and it does not sound good when you say it out loud. But it is the only place in this whole arrangement that I am still allowed to move anything.
评论
0 条评论