Trend · rank 208 of 250

Skilled & research talent migration

Highly skilled workers and researchers increasingly cluster in a few attractive labour markets and research centres, redistributing national expertise and weakening source countries that cannot replace or draw back talent.

Human CapitalPeople, health & workwork skillsdemographygeoeconomics

Second pass: Carried forward from the first pass, where it was Brain drain & talent migration. Held by all three panel models.

Sensitivity and network position

Rank
208of 250
Sensitivity
7.5rank 158
Breadth
1of 20 scenarios High or above
PageRank
20.3rank 205
Eigenvector
1.0
Betweenness
1.3
Links
7trends at Moderate similarity or above
Composite
12.2

Attention over three years

How much this trend is read about on Wikipedia and searched for on Google, week by week, set against the typical trend in the list so that what moves every trend at once is taken out. 100 is this trend's own three-year average. The score measures public attention, which is not the same as the strength of the force: a trend can deepen unnoticed, and a news event can lift attention for a fortnight.

Now
79
Over a year
−36%Fading
Over three years
−27%
Peak
27714 Sept 2025
Sources
Bothagreement 0.72
Coverage
17languages · 55 nations

Trailing 4-week average. 100 is this trend’s three-year average.

By language

Wikipedia reading in each language edition where the trend’s pages are read enough to measure (17 of 28). Share is of all the readers counted.

LanguageShareNowOver a yearThree years
English80%85−31%
Russian4%104+19%
German3%98−18%
Turkish2%37−26%
Chinese2%105−42%
Spanish2%97+3%
French1%73−17%
Arabic1%55+15%
Persian1%82−12%
Vietnamese1%100+53%
Italian1%70+1%
Japanese1%107+1%
Polish1%82+22%
Portuguese1%95−1%
Thai0%75+104%
Hebrew0%162+191%
Dutch0%84+37%

By country

Google search interest over the three years, as a share of each country’s searching. 100 is the nation where interest is highest. The search terms are English, which favours countries that search in English.

United Kingdom

Jamaica

Nepal

Nigeria

Ghana

Qatar

Sri Lanka

United States

Singapore

Ethiopia

United Arab Emirates

Pakistan

100

65

58

53

50

37

36

36

35

34

33

32

The 12 highest of 55 nations with a reading.

By region and G20 member

Attention to this trend inside each place, from Google searches made in its countries and Wikipedia reading in the languages mostly read there, averaged where both exist. Each line compares the trend with its own past in that place; the lines do not compare places.

Regions

PlaceRead fromNow90 days1 year3 yearsThree years
North AmericaGoogle in 2 countries + Wikipedia in English77+7%−56%−16%
EuropeGoogle in 7 countries + Wikipedia in German, French, Italian, Polish, Dutch100+18%−11%−12%
ChinaGoogle in 1 country89+36%−20%−22%
Indo-PacificGoogle in 10 countries + Wikipedia in Japanese, Chinese, Vietnamese, Thai97+9%−17%+1%
South AsiaGoogle in 4 countries66+42%−70%−42%
Gulf and Middle EastGoogle in 2 countries + Wikipedia in Persian, Arabic, Turkish, Hebrew66+14%−43%−45%
AfricaGoogle in 3 countries63−16%−40%−62%
Latin America and CaribbeanGoogle in 5 countries + Wikipedia in Spanish, Portuguese89+8%−13%−21%
Russia and EurasiaWikipedia in Russian104+7%+19%+1%

G20 members

PlaceRead fromNow90 days1 year3 yearsThree years
ArgentinaGoogle50+19%−39%−61%
AustraliaGoogle83+19%−42%−23%
BrazilGoogle + Wikipedia in Portuguese89+10%−7%−31%
CanadaGoogle56+10%−80%−28%
ChinaGoogle89+36%−20%−22%
FranceGoogle + Wikipedia in French83+22%−24%−34%
GermanyGoogle + Wikipedia in German96+2%−29%+9%
IndiaGoogle64+29%−78%−27%
IndonesiaGoogle60−44%−34%−32%
ItalyGoogle + Wikipedia in Italian99+11%+6%−23%
JapanGoogle + Wikipedia in Japanese101+11%−8%+0%
South KoreaGoogle148+169%+8%+78%
MexicoGoogle98+29%−27%+2%
RussiaWikipedia in Russian104+7%+19%+1%
South AfricaGoogle76−46%−31%−52%
TürkiyeGoogle + Wikipedia in Turkish58−2%−30%−50%
United KingdomGoogle63+11%−61%−47%
United StatesGoogle + Wikipedia in English79+12%−52%−15%

Relation to each scenario

A trend relates to a scenario when it changes how likely the scenario is, how hard it lands, or is itself sharply changed by it. Each score is the median of three scores given separately by the members of the model panel.

RelationScenarioFamily
4 · HighAI-directed research produces a reproducible discovery step S17 ↗Technology and infrastructure
2 · LowA cross-regional market access compact redirects investment S12 ↗Economic and trade
2 · LowCheap firm clean power becomes available at scale S15 ↗Energy, climate and resources
1 · Very lowAn AI deployment step causes abrupt labour displacement S08Technology and infrastructure
0 · Not relatedCritical trade divides into rival blocs S01Economic and trade
0 · Not relatedA global funding seizure reaches sovereign balance sheets S02Economic and trade
0 · Not relatedA regional war closes a globally important sea route S03Geopolitical and security
0 · Not relatedHybrid coercion ends in an infrastructure blackout S04Geopolitical and security
0 · Not relatedA disputed transfer of power breaks effective government S05Geopolitical and security
0 · Not relatedTwo breadbasket failures trigger a food availability crisis S06Energy, climate and resources
0 · Not relatedA major disaster disables the national economic core S07Energy, climate and resources
0 · Not relatedA cryptographic breakthrough invalidates digital trust S09Technology and infrastructure
0 · Not relatedA novel respiratory pathogen exceeds health surge capacity S10Health
0 · Not relatedA creditor agreement releases fiscal capacity S11 ↗Economic and trade
0 · Not relatedA great-power agreement lowers economic security barriers S13 ↗Geopolitical and security
0 · Not relatedA verified Middle East settlement restores access S14 ↗Geopolitical and security
0 · Not relatedA validated farming package raises water-efficient output S16 ↗Energy, climate and resources
0 · Not relatedA public service platform reaches inclusive operation S18 ↗Technology and infrastructure
0 · Not relatedAffordable preventive treatments reduce chronic illness S19 ↗Health
0 · Not relatedA live outbreak proves rapid distributed pandemic defence S20 ↗Health

Similarity = 0.3 × shared tags (Jaccard) + 0.3 × likeness of scenario profiles (cosine) + 0.3 × likeness of names and descriptions (TF-IDF cosine) + 0.1 if in the same capability domain, cut into seven levels at 0.12, 0.22, 0.33, 0.45, 0.58, 0.72. Shown at Moderate and above.

SimilarityTrendThemeRank
4 · HighTechnical & occupational skill shortagesPeople, health & work19
3 · ModerateAI & knowledge-work automationPeople, health & work53
3 · ModerateAutomated AI researchAI, compute & quantum83
3 · ModerateLifelong learning & digital reskillingPeople, health & work132
3 · ModerateYouth population expansionPeople, health & work146
3 · ModerateYouth employment exclusionPeople, health & work177
3 · ModerateGraduate skill mismatchPeople, health & work178