Skip to main content

Those Who Leave and Those Who Arrive

The total recovers, the make-up changes completely: why no one can say why

@Mythogen Engine · Oct 23, 2026

“Snatching Rice from the Beggar's Bowl”, Part 4 | Previous: “The Fraud You Can See, the Numbers Nobody Measures” | Next: Final part “After the Counting”

Not resenting the rich, not hating the poor: struggle is not competition, and in the end it reaches every floor.

We only count the net​

On his first day at the Census and Statistics Department, the new recruit is handed a calculator by his supervisor. “Every day some people leave and some people arrive. You keep the record.”

The recruit asks: “Which column do I put the leavers in, and which column for the arrivals?”

The supervisor says: “No need to separate them. Press minus for each one who leaves and plus for each one who arrives, and report the net at the end of the month.”

“Then how will we know who left and who came?”

The supervisor thinks for a moment. “If the net is positive, we call it a population recovery.”

This is a fictional fable, but every detail has a basis, and each is set out below. A real example is close at hand: a friend lives in a private estate completed in 2012. Most of the original neighbours were local families; in recent years they have moved out one after another, and most of those moving in are new arrivals from the mainland. This is not public housing, and it has nothing to do with welfare: the change is happening at every level of society. Many people around me feel the same, and feel it strongly; but ask how many, and no one can produce a direct number.

The first three essays in this series asked about distribution: who was shortchanged, what was miscounted. This one asks a question further upstream: how the calculator itself is designed.

1. More than one way in​

What people usually call “new immigrants” is only one point of entry; people who came through the others have never been counted as new immigrants.

RouteStartScaleCounted as “new arrivals”?
One-Way Permit150 places a dayUp to about 55,000 a yearYes
Children of two non-resident parents (“double non”)After the 2001 Chong Fung-yuen ruling202,314 (official births, 2001–2012)No: permanent residents at birth
Admission Scheme for Mainland Talents and Professionals (ASMTP)July 2003About 83,700 approvals in 2003–2015; about 12,000–14,000 a year in 2017–2019No
Quality Migrant Admission Scheme (QMAS)June 2006About 3,000 in 2006–2015; quota raised from 1,000 to 2,000 a year from 2020No
Immigration Arrangements for Non-local Graduates (IANG)May 2008About 51,500 in 2008–2015; about 10,000 a year in 2017–2019No
General Employment Policy (not for mainland residents)Before 2006About 273,100 in 2006–2015; about 40,000 a year in 2017–2019No
Top Talent Pass Scheme (TTPS)December 2022About 105,000 arrived (end-2025)No
All talent schemes combined—End-2022 to end-2025: over 260,000 talents and over 240,000 dependantsNo

Sources: One-Way Permits from a LegCo paper, 2014; “double non” births from a Department of Justice paper to LegCo, 2013; admission schemes up to 2015 from Audit Commission Report No. 66, Chapter 4; 2017–2021 from a LegCo reply, September 2021; TTPS and totals from Sing Tao.

  • Approval does not rest with Hong Kong: for One-Way Permits, “receipt, approval and issue fall within the remit of the mainland authorities”, while the SAR government “cooperates at the case level”; the quota is set by the central government after consulting the SAR under Article 22 of the Basic Law.
  • What “new immigrant” means: officially, “new arrivals” are “One-Way Permit holders who have resided in Hong Kong for less than seven years” (LegCo Research Office, 2017). Talents and dependants over three years exceed 500,000, about three times the three-year One-Way Permit cap (about 164,000), yet have never been counted as new immigrants.
  • TTPS changed the scale: approvals across all talent schemes rose from 38,559 in 2022 to 138,215 in 2024 (JLL, citing Immigration Department data).
  • Shortcuts to seven years: the main applicant must live here for seven years, but (1) mainland study agents widely advertise that full-time study counts toward the seven years, so a four-year degree plus three years on IANG is enough (not publicly confirmed by the Immigration Department); (2) children of Chinese citizens born in Hong Kong are permanent residents at birth; (3) spouses and children count their years together with the main applicant.
  • How “double non” came about: on 20 July 2001, the Court of Final Appeal ruled in the Chong Fung-yuen case that Chinese citizens born in Hong Kong are permanent residents regardless of their parents' status. Babies born in Hong Kong to mainland women whose husbands were not permanent residents rose from 620 in 2001 to a peak of 35,736 in 2011, totalling 202,314 from 2001 to 2012; the government imposed a “zero quota” from 2013 (Department of Justice paper to LegCo, May 2013; LegCo paper, 2002).

2. So-called talent​

“Talent” is defined by criteria on paper, not by what people do in Hong Kong.

  • Three TTPS categories: Category A, annual income of HK$2.5 million or more; Category B, a bachelor's degree from a global top-100 university plus three years' work experience; Category C, a top-100 graduate with under three years' experience, subject to a quota. None requires a job in Hong Kong.
  • Origin: over three years, all talent schemes approved more than 410,000 applications, about 73% (about 300,000) from the mainland; about 95% of TTPS applicants are from the mainland (Secretary for Labour and Welfare's reply, March 2026, HK01).
  • What they do once here: a Labour and Welfare Bureau survey (of those here six months, response rate close to 40%) found about 54% employed, with median monthly income of about HK$50,000 (LegCo paper, June 2025). What the other 46% are doing, there is no figure; a 40% response rate may also overstate income, and the government's claim that talents “contribute 1.2% of GDP a year” rests on this very survey.
  • Renewals: the TTPS renewal rate is about 52% (Sing Tao).
  • Housing: a government survey found 70% renting private flats; JLL estimates this brings about 12,000 units of net rental demand a year (JLL).
  • Fraud: in 2025 the Immigration Department broke up a syndicate that used fake degrees and fake employment proof to help at least 22 TTPS applications, arresting 18 people; only from June 2024 did Categories B and C have to submit verified degrees (Immigration Department). How many problematic applications were approved before that, nobody knows.
  • Children: from January to October 2023, seven talent schemes brought in 47,493 dependants under 18, who may attend subsidised schools; the Education Bureau “does not keep data on dependants studying in Hong Kong” (Sing Tao). University fee subsidies have since been tightened: from the 2028 JUPAS round, children of talents must have lived in Hong Kong for two years (Sing Tao).
  • Welfare: the Social Welfare Department excludes people “subject to a condition of stay under regulation 2 of the Immigration Regulations”, with visitors and imported workers as examples; the Immigration Department says TTPS holders, including after renewal, are “not subject to any other condition of stay” (Immigration Department TTPS FAQ). Whether a TTPS holder who has lived here a year can apply for CSSA is not made clear by either document; the Social Welfare Department would need to answer directly.

Whether talents end up using public housing, welfare, school places and healthcare, nobody knows; once they have lived here seven years, nobody ever will.

3. Invisible once inside​

Once people become permanent residents, every statistic counts them as “local”, and their route in disappears.

RouteBecomes a permanent residentCounted afterwards as
Children of one or two non-resident parentsAt birthLocal
One-Way PermitAfter seven yearsLocal
Talent schemesAfter seven yearsLocal

Hall places at the eight publicly funded universities (at the start of the academic year; University Grants Committee reply, November 2025):

Academic yearLocal studentsNon-local studentsTotalNon-local share
2024/2517,45321,16838,62154.8%
2015/1619,39515,11434,50943.8%

Over ten years the total rose by 4,112 places, while places for local students fell by 1,942 (−10%). The University Grants Committee says it “does not keep data on hall places by students' place of origin”.

  • Non-local ≠ mainland: in 2022/23, of about 22,000 non-local students on UGC-funded programmes, about 75% came from the mainland (RFA, citing C&SD); hall allocation is not recorded by place of origin.
  • Local ≠ grew up in Hong Kong: from 2001 to 2012, 202,314 children whose fathers were not permanent residents were born in Hong Kong; they are now reaching university age, and statistically they are all local students. They had no choice; but no one planned for the school places, healthcare and housing they would need.
  • The universities' reason: non-local students have no home in Hong Kong, so universities have reason to give them priority; allocation rules are set by each university (government reply to LegCo, 2024). The ratio alone does not prove unfairness, but it does prove that no one is counting.

4. Only the net is published​

The Census and Statistics Department lists only One-Way Permit holders separately; all other inflows and outflows are merged into one “net” figure.

PeriodOne-Way Permit arrivalsNet movement of other residentsSource
End-2024 to end-2025About 25,000Net inflow about 4,000HK01
Mid-2024 to mid-2025About 32,000Net outflow about 13,800 (provisional)World Journal
Mid-2022 to mid-2023About 31,000Net inflow about 143,000Human Resources Online
Mid-2021 to mid-2022About 18,300Net outflow about 113,200The Standard

This essay could not verify the breakdown for mid-2023 to mid-2024 from published reports.

  • Those who left: excluding One-Way Permit arrivals, Hong Kong residents saw a net outflow of about 49,100 in 2020 and about 69,200 in 2021, against a net inflow of about 5,200 in 2019 (HKCSS). By June 2025 the UK had granted 181,609 BN(O) visas, and over 166,300 people had arrived in Britain (Hong Kong Watch, citing the Home Office). A 2021 UK government survey found that 69% of BN(O) holders had a bachelor's degree or above and 39% worked in professional jobs.
  • Those who came: talent schemes at the top, One-Way Permits at the bottom. Those leaving were mainly middle-class professionals; those arriving fill in at both ends.
  • No direct emigration statistics: the Security Bureau “does not keep direct statistics”, estimating only from applications for certificates of no criminal conviction, and admits that applicants do not necessarily emigrate (HKFP, 2018).
  • Direction given, breakdown withheld: the government says population growth comes from talents and imported workers “offsetting natural decrease”, but does not publish how many talents arrived or how many local residents left.
  • Even the total gets revised: the mid-2025 population was first announced as “three years of growth in a row”, then revised to a 0.3% year-on-year fall.

The word “net” has already cancelled out locals leaving against talents arriving. The calculator in the fable, with only plus and minus keys, works exactly this way.

5. Taiwan can do it​

Taiwan also has a large incoming population, and publishes the numbers monthly by nationality; “cannot be measured” is not a technical problem.

CategoryWhat Taiwan publishesHong Kong's equivalent
Foreign workersOver 860,000 industrial and care migrant workers: Indonesia 327,976, Vietnam 291,540, Philippines 175,786, Thailand 70,158 (February 2026, TVBS citing the Ministry of Labor)Imported workers and talents merged into “net movement of other residents”
Marriage migrantsThe National Immigration Agency's table of foreign and mainland (incl. HK and Macau) spouses, updated continuously from January 1987 to August 2026, by original nationality (National Immigration Agency)One-Way Permits are counted, but untraceable after seven years
Professionals12,082 Employment Gold Cards issued by end-2024, with nationality shares published: US 25%, Hong Kong 12%, Japan 8% (TechNews citing the National Development Council)Talent schemes report approvals; on whether people stay, the Immigration Department has “no regular compilation”

Taiwan's numbers have blind spots too, such as the gap between the official price index and “felt prices” (see Part 2). But on “who came, and from where”, Taiwan has shown that a society can publish numbers monthly and by nationality, and the sky does not fall.

6. What people feel, and the loopholes​

For each loophole, one must first separate “possible under the rules” from “how many people actually do it”; the first can be verified, the second mostly has no numbers.

LoopholeCheckable sourceRuleScale
Children of one or two non-resident parents count toward “half the household has lived here seven years” for public housingHousing Authority application guide (how Hong Kong-born children count)Verified: since 2005, children under 18 born in Hong Kong with established permanent residency are deemed to meet the seven-year rule, regardless of their parents' years of residence or immigration status (LegCo paper)No figures
Bringing parents to Hong Kong through childrenOne-Way Permit categories, LegCo papersCategory verified (elderly dependants aged 60+)No published figures
Divorce and remarriage, sham marriagesImmigration Department prosecution figures, court judgmentsVerified: a criminal offence; false statements carry up to 14 years in prisonOnly individual cases: in November 2025 a syndicate was broken that had arranged over 60 sham marriages in 2023–2024, with 37 arrests (Immigration Department)
Fake degrees in talent applicationsImmigration Department press releasesVerified: no verified documents required before June 2024Only the 22 cases uncovered
Study years counting toward seven yearsImmigration DepartmentNo public statement from the Immigration Department; only study agents' claimsNo figures

7. Asking for data is not discrimination​

No data, plus “don't discriminate”, makes a complete gag order: the system doesn't record, so there is no data; without data, people can only talk about what they feel; once they talk about feelings, they are accused of discrimination; so no one asks for data any more.

Whenever someone asks “do these people live in public housing or claim welfare?”, the answer is usually not a number but “don't divide, don't discriminate”. But asking for data is not discrimination. Discrimination is treating someone differently because of where they come from; accountability is asking the government to account for the consequences of each policy.

What really pushes suspicion into discrimination is refusing to publish data: the information vacuum fills with rumour, and every newcomer falls under suspicion, including the honest majority. Publishing public housing, welfare and school-place figures by entry route protects precisely them. The demand applies equally to One-Way Permits, “double non” children, every talent scheme and overseas talent; Taiwan publishes figures on migrant workers and new residents by nationality, and no one calls that discrimination.

This is the same move as the one criticised in Part 2 of this series, on Factfulness: “the world is getting better” is used to tell people not to complain, and “don't discriminate” is used to tell people not to ask. Both turn a structural problem into a flaw in the questioner's character.

8. “Same language, same people”​

When the population is shrinking and wages seem too high, the quickest answer is to bring people in. But immigration itself is not a tool for solving distribution. France's banlieues show the other side: generations living in the same housing estates, unemployment persistently above the national rate, and clashes with police that twice, in 2005 and 2023, escalated into nationwide riots. The problem is not where the migrants came from, but that after bringing people in, governments did not invest at the same pace in housing, education and jobs, and left the cost of integration to the poorest communities (on how the banlieues came to be, see section 12 of Part 1).

Hong Kong's situation is often waved through with the phrase “same language, same people”. But 150 years of British rule left behind not just English but a whole set of institutional habits: common law, due process, expectations about public rules; post-reform China took a different road. The two places do not necessarily speak the same Chinese, write the same script, or believe in the same rules.

Yet official discourse does not even acknowledge the difference: senior-secondary Citizenship and Social Development textbooks from 2022 state that Hong Kong was not a colony, only under British “colonial rule” (NPR). If the difference does not exist, integration cannot become a policy issue; and if it is “same language, same people, no integration needed”, then naturally there is neither measurement nor investment. The cost of integration falls, again, on the poorest communities: the local poor and the newly arrived poor are set in front of the same bowl to fight over the rice.

Conclusion: not that it can't be measured, but that no one measures​

Six departments, one answer: not kept.

DepartmentWhat it does not measure
Immigration DepartmentHow many talent entrants obtain right of abode and how long they stay; the Audit Commission said in 2016 that the computer system “could not readily generate” this (Audit Commission Report No. 66, Chapter 4)
Security BureauEmigration numbers, estimated only from certificates of no criminal conviction
Census and Statistics DepartmentLocal residents leaving and new residents arriving, merged into one net figure
University Grants CommitteeHall places by place of origin
Education BureauTalent dependants studying in Hong Kong
Housing AuthorityOffshore and mainland assets (see Part 3)

The data exists: the census records place of birth and years of residence, the Immigration Department holds a record of every visa, and Taiwan has shown that publishing by nationality can be done. The Audit Commission pointed out in 2016 that the Immigration Department did not compile statistics on talent entrants obtaining right of abode, and this essay found no figures published since; the Census and Statistics Department once published a thematic report on “persons from the mainland having resided in Hong Kong for less than seven years”, but the one this essay could find still uses figures from the 2006 by-census. To answer this essay's questions, only a few numbers are needed: how many people under each talent scheme have become permanent residents; how many talent dependants attend subsidised schools; whether full-time study counts toward the seven years. Anyone can request them under the Code on Access to Information; if the reply is “not kept”, that is itself the answer. It is not that the numbers cannot be measured, but that no one measures; and if no one measures, no one has to take responsibility.

The four essays of “Snatching Rice from the Beggar's Bowl” say the same thing. Part 1: people who don't work are treated as the problem, while the gains from rising productivity never reached working people. Part 2: the ranking tells people to look down, while the richest 1% of the world owns about 37% of its wealth. Part 3: the means test measures only payslips, while no one asks who collects the land rent or why the treasury depends on land sales. Part 4: statistics report only net figures; talent schemes bring about 12,000 units of rental demand a year, and the gains go to those who collect rent; how many locals left, and who moved into whose flat, no one counts.

In all four places, someone is told to stare at people poorer than themselves: welfare recipients, children in Africa, public housing applicants, new arrivals. Look at any YouTube comment section: the voices treading downward are far louder than the voices questioning the top. Yet even if you trample everyone poorer than you, your own situation will not change; the money is still at the top. An old Chinese saying goes, “Do not worry about scarcity; worry about inequality.” Today's problem is exactly inequality, and policy keeps it in place. This is not resenting the rich, nor hating the poor. Competition means each relies on their own ability; struggle means each floor treads on the one below, and in the end it reaches every floor, including those who think they stand high enough. What needs doing is to see the structure clearly, and to take back the share that is due.

When statistics are designed so that a problem can never appear, “no evidence” is no longer an answer but part of the problem. The burden of proof should rest on those who make the decisions, not on those who pay the rent and stand in the queue.

“Snatching Rice from the Beggar's Bowl” series: Prologue | Part 1 “Suffer If You Don't Work?” | Part 2 “Global Top, Local Bottom” | Part 3 “The Fraud You Can See, the Numbers Nobody Measures” | Part 4 “Those Who Leave and Those Who Arrive” | Final part “After the Counting”

Sources​

Most sources are in Chinese; titles are translated.