Regional Review 2026

SkyMind Regional Review 2026

An annual descriptive review of regional change based exclusively on official public statistics. 32 countries, 1,948 regions, up to 48 metrics per country, 2019–2023.

Published 1 July 2026· 32 countries· 1,948 regions· ~20 min read
Contents

1. What We Observed

Seven observations from the data. Each can be verified from the published dataset at sky-mind.com.

  1. A natural disaster is visible in regional statistics within one reporting cycle. Three Turkish provinces hit by the February 2023 earthquake dropped to the bottom of the national ranking — despite nominal GDP growth. Life expectancy in Hatay fell by 24 years in a single data period.
  2. Regions can improve their composite position while losing population. In Portugal, Poland, Greece, Moldova, and Italy, regions that lost 2–11% of their population between 2019 and 2023 still moved up in their national rankings. Demographic decline alone did not determine composite trajectory.
  3. Employment recovery reshaped rankings more than tourism decline. In Greece, western mainland regions saw unemployment fall from 24% to 10% between 2019 and 2023. The Ionian Islands did not decline — they were overtaken by regions that had started from deeper crisis.
  4. GDP can double without creating proportional employment — but it can reverse emigration. Across Bulgaria, GDP per capita grew 55–115% while employment barely moved. Yet 22 of 28 regions flipped from net emigration to net immigration. Economic convergence changed migration patterns before it changed labor markets.
  5. The automotive transition is already measurable at district level. Wolfsburg — Germany's highest GDP-per-capita district — saw GDP decline for the first time in the dataset. Meanwhile, Böblingen (Mercedes heartland) grew manufacturing GVA by 59%. The EV transition is splitting Germany's industrial belt.
  6. Rapid demographic aging had no measurable short-term effect on composite rank. Regions that aged fastest (>2 years of median age increase in 4 years) showed the same average rank change as those that aged slowly. The economic catch-up in the Balkans and southern periphery offset the demographic headwind.
  7. Some regions have held the same rank for a decade. München has been #1 out of 401 German districts for every year in the dataset. &İstanbul, Stockholm, Helsinki, and nine other regions show zero rank variation. At the bottom, Karlovarský kraj (Czech Republic) and Ticino (Switzerland) have been last in their countries for every recorded year.

2. What Changed

Five-year rank change, 2019–2023. Composite rank within each country, based on min-max normalized economic and demographic metrics from Eurostat and national statistical offices.

Methodological note

Composite scores use min-max normalization per country and year. A score of 60 in 2019 and 60 in 2023 do not represent the same absolute position — the scale endpoints change each year. For this reason, all cross-year comparisons in this review use rank within country, not composite score.

The comparison window is 2019–2023. Year 2024 was excluded because most Eurostat economic indicators have not yet been published for 2024 at NUTS-3 level. Including 2024 would reduce the economic metric basis from 18–23 indicators to 3–4.

Census corrections. Germany conducted a major census in 2022 (Zensus 2022), which revised population figures downward in several cities. Albanian regions show extreme migration rates in 2023 due to a census that revealed population was ~400,000 lower than estimates. Rank changes in demographic-heavy metrics should be interpreted with caution.

Rank gain = positions moved up (positive = improvement). gain_pct normalizes for country size: rank change / total regions × 100.

Turkey (81 provinces)

Region20192023ChangeDriver
Çankırı8121+60Relative gain as earthquake regions collapsed
Siirt7314+59
Şırnak7522+53
Adıyaman3179−48Feb 2023 earthquake: death rate ×4, construction +509%
Kahramanmaraş3380−47Feb 2023 earthquake: life expectancy −24 years
Hatay3781−44Feb 2023 earthquake: net migration −83‰

Germany (401 districts)

Region20192023ChangeDriver
Ilm-Kreis (Thuringia)330136+194East German convergence
Saalfeld-Rudolstadt (Thuringia)359181+178East German convergence
Wartburgkreis (Thuringia)318142+176East German convergence
Salzlandkreis159333−174
Saalekreis116289−173
Wittenberg226397−171

Part of the German rank movement may reflect Zensus 2022 population corrections rather than real demographic change.

Greece (52 regions)

Region20192023ChangeDriver
Ileia4222+20Unemployment 24% → 10%
Aitoloakarnania3824+14Unemployment 24% → 10%
Fthiotida4533+12
Kerkyra (Corfu)2343−20Stagnant employment, overtaken
Ithaki, Kefallinia2646−20Stagnant employment, overtaken
Lefkada3248−16Stagnant employment, overtaken

Portugal (25 regions)

Region20192023ChangeDriver
Alto Tâmega2511+14Interior improvement
Terras de Trás-os-Montes2210+12
Alto Minho186+12
Alentejo Central617−11
Baixo Alentejo515−10
Alentejo Litoral412−8

Bulgaria (28 regions)

Region20192023ChangeDriver
Kyustendil2815+13GDP convergence
Vidin2717+10GDP convergence
Lovech2416+8GDP convergence
Dobrich515−10
Plovdiv38−5
Burgas611−5

Croatia (21 counties)

Region20192023ChangeDriver
Sisačko-moslavačka155+10Post-2020 earthquake reconstruction
Koprivničko-križevačka169+7Construction GVA +297%
Karlovačka1711+6Construction GVA +169%
Virovitičko-podravska715−8
Osječko-baranjska814−6
Brodsko-posavska1821−3

Serbia (25 regions)

Region20192023ChangeDriver
Borska oblast144+10GDP per capita tripled (€8.3k → €23.9k)
Toplička oblast2115+6
Pirotska oblast2217+5

Finland (19 regions)

Region20192023ChangeDriver
Etelä-Savo189+9Eastern Finland convergence
Pohjois-Karjala168+8Eastern Finland convergence
Pohjois-Savo157+8Eastern Finland convergence
Kymenlaakso917−8Southern industrial decline
Päijät-Häme815−7Southern industrial decline
Varsinais-Suomi710−3

Eastern Finland — traditionally the country's poorest region — moved into the top half. Southern industrial regions moved from the top half to the bottom.

Italy (107 provinces)

Region20192023ChangeDriver
Genova6549+16
Sassari8874+14
Ragusa6854+14
Frosinone4964−15
Grosseto5567−12
Chieti6374−11

Poland (73 subregions)

Region20192023ChangeDriver
Łomżyński4529+16
Bialski7158+13
Szczecinecko-pyrzycki6755+12
Piotrkowski2642−16
Bydgosko-toruński3248−16
Oświęcimski2439−15

Romania (42 counties)

Region20192023ChangeDriver
Sălaj2113+8
Harghita2518+7
Alba2217+5
Vaslui721−14Largest decline in Romania
Bacău916−7
Botoşani1724−7

Norway (12 regions)

Region20192023ChangeDriver
Trøndelag51+4Overtook Oslo as national leader
Troms og Finnmark63+3Northern Norway rising
Innlandet119+2
Viken48−4
Oslo14−3Lost national leadership
Vestland35−2

Oslo lost the #1 position for the first time in the dataset. Trøndelag (Trondheim region) rose from 5th to 1st.

Netherlands (40 regions)

Region20192023ChangeDriver
Overig Groningen3223+9Post-gas-extraction transition
Overig Zeeland1814+4
Achterhoek2117+4
Noord-Friesland1519−4
Agglomeratie Haarlem26−4
Zuidwest-Drenthe2427−3

3. Observed Regional Patterns

Eleven patterns from 2019–2023 data. Each follows a fixed format: Pattern / Observation / Evidence / Why it matters / How to verify.

Pattern 1

Reconstruction without recovery

Observation

Three Turkish provinces hit by the February 2023 earthquake dropped to the bottom of the national ranking despite nominal economic growth. The composite captured a signal that GDP alone missed: economic reconstruction was underway, but the demographic damage was severe enough to dominate the overall position.

Evidence

Hatay fell from rank 37 to 81 (last of 81 provinces). Its construction GVA rose 226% and GDP per capita increased 72% — both consistent with large-scale reconstruction spending. But life expectancy dropped from 79.0 to 55.4 years (−24 years), the death rate quadrupled from 4.7 to 17.1 per thousand, and net migration reached −82.9 per thousand. Kahramanmaraş (rank 33 → 80) and Adıyaman (31 → 79) show the same split: economic metrics up, demographic metrics catastrophic.

Why it matters

Economic indicators can mask demographic devastation. A region can appear to be “growing” by standard output metrics while experiencing the worst demographic shock in its recorded history. Composite indices that include both economic and demographic dimensions capture this divergence; single-axis indicators do not.

How to verify

sky-mind.com/map.html → Turkey → compare GDP per capita (above median for all three) with Life expectancy (lowest three in the country) for Hatay, Kahramanmaraş, Adıyaman.

Pattern 2

Employment normalization, not tourism decline

Observation

All four Ionian Island regions in Greece (Kerkyra, Kefallinia, Lefkada, Zakynthos) dropped 15–20 positions between 2019 and 2023. But they did not decline in absolute terms — they were overtaken by western mainland regions whose labor markets normalized after the crisis.

Evidence

Aitoloakarnania and Ileia (western Peloponnese) saw unemployment drop from 24.1% to 9.8% (−14.3 pp) and employment rise from 53.2% to 65.9% (+12.7 pp). Meanwhile, the four Ionian islands’ employment barely moved: 64.7% → 65.3% (+0.6 pp), while unemployment rose from 12.4% to 14.7%. The islands’ GDP per capita still grew 19–30%.

Why it matters

The headline “Greek islands are declining” would be incorrect. The islands’ absolute metrics improved or held steady. What changed was that regions starting from much higher unemployment recovered faster — a classic post-crisis convergence effect in labor markets.

How to verify

sky-mind.com/map.html → Greece → switch between Employment rate (mainland high, islands moderate) and GDP per capita (islands still above median).

Pattern 3

Population decline did not predict composite decline

Observation

Across at least five countries, regions that lost 2–11% of their population between 2019 and 2023 improved their composite rank. Demographic shrinkage alone was not a sufficient predictor of overall trajectory.

Evidence

RegionCountryPop changeRank change
Alto TâmegaPT−2.3%+14
ŁomżyńskiPL−8.3%+16
AitoloakarnaniaGR−5.2%+14
LeovaMD−11.5%+15
SassariIT−2.5%+14

In each case, economic metrics (GDP, employment, or GVA) improved enough to offset the demographic drag. Alto Tâmega rose from last place in Portugal despite being one of the country’s most depopulated regions.

Why it matters

Population decline is widely treated as an indicator of regional decline. These cases show that the relationship is not automatic: regions can improve their economic profile while shrinking demographically. Whether this represents genuine improvement or simply slower decline relative to peers is a question the data raises but does not answer.

How to verify

sky-mind.com/map.html → Portugal → Alto Tâmega: compare Population total (declining) with GDP per capita (improving).

Pattern 4

Post-earthquake construction is visible across countries

Observation

Regions affected by major earthquakes showed the fastest construction-sector growth in Europe. This pattern appeared independently in Turkey (February 2023 earthquake) and Croatia (March and December 2020 earthquakes), separated by three years and 1,500 km.

Evidence

Turkey: Adıyaman construction GVA +509%, Kahramanmaraş +317%, Hatay +226%. Croatia: Koprivničko-križevačka +297%, Sisačko-moslavačka +282% (the Petrinja/Sisak earthquake zone), Zagrebačka županija +236%. In both countries, construction was the fastest-growing sector in the affected regions.

Why it matters

Infrastructure spending is the fastest-responding economic metric to natural disaster. It inflates GVA and GDP in the affected region, which can create misleading signals in purely economic rankings. Composite indices that include demographic dimensions partially correct for this.

How to verify

sky-mind.com/map.html → Turkey or Croatia → select GVA Construction metric.

Pattern 5

A single industrial investment can reshape a region

Observation

Borska oblast in Serbia saw its GDP per capita nearly triple between 2019 and 2023 — the largest proportional GDP increase of any region in the dataset.

Evidence

GDP per capita rose from €8,300 to €23,900 (+188%). The region moved from rank 14 to rank 4 out of 25 Serbian regions. By comparison, the median Serbian region’s GDP grew approximately 80% over the same period. Borska oblast is the location of the Bor copper mine complex, which received major foreign investment during this period.

Why it matters

Regional statistics can be dominated by a single large employer or project. When one investment changes a region’s GDP by a factor of three, the “region” as a statistical unit becomes a proxy for a single enterprise. This is not a flaw in the data — it is a real feature of small-region economies — but it should inform how regional rankings are interpreted.

How to verify

sky-mind.com/map.html → Serbia → Borska oblast → GDP per capita timeline.

Pattern 6

GDP convergence without labor market change

Observation

Across Bulgaria, GDP per capita grew 55–115% between 2019 and 2023, while employment rates moved less than 2 percentage points. Economic convergence proceeded through productivity gains, not labor market expansion.

Evidence

RegionGDP 2019GDP 2023GDP chgEmpl. chg
Stara Zagora€8,700€18,700+115%+1.3 pp
Sofia (province)€7,500€12,400+65%+0.1 pp
Silistra€4,200€6,700+60%+0.1 pp
Gabrovo€7,900€12,400+57%+0.1 pp

Bulgarian employment rates were already at 74–80% in 2019 — high by European standards. GDP growth came from value-added per worker, not from employing more people.

Why it matters

“Economic growth” and “job creation” are often treated as synonyms. Bulgaria’s case shows they can diverge for structural reasons: when employment is already high, further GDP growth requires productivity improvement, not labor absorption.

How to verify

sky-mind.com/map.html → Bulgaria → compare GDP per capita (large variation) with Employment rate (almost uniform across regions).

Pattern 7

East German districts continued to close the gap

Observation

The three largest rank gainers in Germany (401 districts) were all in Thuringia, a former East German state. Each rose more than 170 positions between 2019 and 2023.

Evidence

Ilm-Kreis rose from rank 330 to 136 (+194), Saalfeld-Rudolstadt from 359 to 181 (+178), and Wartburgkreis from 318 to 142 (+176). All three lost population (−4.6% to −4.8%), meaning the improvement was driven by economic metrics, not demographic growth.

Why it matters

East-West German convergence has been a policy objective since reunification in 1990. These data show the process is ongoing at the district level, with some Thuringian districts moving from the bottom third to the middle in four years. The movement occurred despite population decline.

Caveat

Germany’s 2022 census (Zensus 2022) corrected population figures in many western cities downward. Part of the rank convergence may reflect this statistical correction rather than real economic change.

How to verify

sky-mind.com/map.html → Germany → filter Thuringia → composite score timeline.

Pattern 8

Bulgaria’s migration reversal

Observation

Between 2019 and 2023, 22 of Bulgaria’s 28 regions flipped from net emigration to net immigration. A country that defined European brain drain in the 2010s recorded positive migration across most of its territory by 2023.

Evidence

In 2019, the average Bulgarian region had net migration of −11.8 per thousand. By 2023: +5.2 per thousand — a swing of +17 points. The largest single reversal: Sofia oblast (not the capital) went from −54.6‰ to +3.4‰, a swing of 58 points. Kardzhali shifted from −10.8‰ to +34.5‰, becoming one of Europe’s strongest migration magnets.

This occurred alongside the GDP doubling documented in Pattern 6. Population still declined in absolute terms (the base was smaller), but the emigration hemorrhage stopped.

Why it matters

Migration reversal is among the hardest demographic indicators to move. That 22 of 28 regions achieved it simultaneously suggests a country-level structural shift, not isolated regional success. Whether the migration is return migration, immigration from third countries, or statistical artifact of EU free movement deserves investigation beyond what this dataset can answer.

How to verify

sky-mind.com/map.html → Bulgaria → Net migration metric → compare 2019 and 2023 values across regions.

Pattern 9

Germany’s manufacturing split — diversifiers rose, monocultures stalled

Observation

Germany’s manufacturing heartland split in two between 2019 and 2023. Districts that grew their manufacturing base climbed the ranking. Districts dependent on a single large employer — particularly in automotive — stagnated or declined.

Evidence

Diversifiers (manufacturing GVA grew):

DistrictManuf shareGDP chgRank chg
Olpe (NRW, metalworking)45.6% → 52.8%+29%+54
Böblingen (BW, Mercedes)49.4% → 56.7%+34%+1
Erlangen (Siemens HQ)48.3% → 51.3%+13%+7

Monocultures (manufacturing GVA stalled or fell):

DistrictManuf shareGDP chgRank chg
Wolfsburg (VW HQ)77.3% → 74.3%−4.3%−4
Salzgitter (steel)61.3% → 53.7%+1%
Emden (VW plant)50.0% → 44.5%−33

Wolfsburg — Germany’s highest GDP-per-capita district — saw GDP decline from €193,400 to €185,100. Its manufacturing GVA was flat (€16.8B → €16.2B) while Böblingen’s grew 59% (€11.1B → €17.7B).

Why it matters

The automotive transition is Europe’s largest industrial transformation. These data show that its effects are already measurable at regional level — and that the impact is not uniform across automotive regions. Districts with diversified manufacturing outperformed single-employer districts.

How to verify

sky-mind.com/map.html → Germany → Wolfsburg → GDP per capita timeline (declining); compare with Böblingen (rising).

Pattern 10

Rapid aging had no measurable effect on composite rank

Observation

Regions that aged fastest (median age increase >2 years in 4 years) showed virtually the same average rank change as regions that aged slowly or rejuvenated.

Evidence

Of 1,187 regions with data for both years, 104 aged by more than 2 years in median age. Their average rank change: −0.4 (essentially zero). The remaining 1,083 regions averaged +0.2. North Macedonia is the cleanest example: all 8 regions aged by 2.4–4.2 years (Istocen: 41.9 → 46.1), yet the national ranking was unchanged — Skopje stayed #1, the median rank shift was zero.

Why it matters

Aging is widely cited as a long-term structural risk for regional economies. These data do not contradict that view — four years may be too short to see the economic effects. But they do show that rapid aging did not cause measurable short-term composite decline.

How to verify

sky-mind.com/map.html → North Macedonia → compare Median age (high everywhere) with composite score (no systematic decline).

Pattern 11

Finland’s internal geography reversed

Observation

Eastern Finland — traditionally the country’s poorest macro-region — moved from the bottom third to the top half of the national ranking between 2019 and 2023. Southern industrial regions moved in the opposite direction. Finland’s internal economic geography inverted.

Evidence

Risers (eastern Finland):

Region20192023ChgUnemployment
Etelä-Savo189+97.8% → 6.9%
Pohjois-Karjala168+87.8% → 6.9%
Pohjois-Savo157+87.8% → 6.9%

Decliners (southern industrial):

Region20192023ChgUnemployment
Kymenlaakso917−86.3% → 7.5%
Päijät-Häme815−76.3% → 7.5%

Long-term unemployment in the southern industrial regions doubled from 1.1% to 2.3%, while in eastern Finland it fell from 1.2% to 1.1%. Both groups lost population, so the divergence was driven entirely by labor market dynamics.

Why it matters

Finland’s east-west divide has been a persistent feature of its economic geography since industrialization. These data suggest the divide may be narrowing — or even reversing — at NUTS-3 level. Whether this reflects structural change (remote work, public-sector stability in regional capitals) or a cyclical effect of post-COVID industrial adjustment is a question the data raises but cannot answer.

Caveat

Employment and unemployment data for Finnish NUTS-3 regions are pooled at NUTS-2 level by Eurostat. All three eastern regions share identical values, as do both southern regions. The pattern is real at NUTS-2 level but cannot be differentiated within each NUTS-2 area.

How to verify

sky-mind.com/map.html → Finland → compare Unemployment rate (east improving, south worsening) with composite score rankings.


4. Persistent Stability

Regions whose rank did not change across the full available time series (2014–2023 for most countries). Stability is measured as the standard deviation of annual rank: zero means the region held exactly the same position every year.

Permanent leaders

RegionCountryRankDurationTotal regions
MünchenDE#110 years401
&İstanbulTR#111 years81
Stockholms länSE#111 years21
Helsinki-UusimaaFI#110 years19
Kyiv CityUA#110 years25
Hlavní město PrahaCZ#111 years14
Grad ZagrebHR#111 years21
City of BelgradeRS#18 years25
Chişinău MunicipalityMD#111 years35
Osrednjeslovenska (Ljubljana)SI#111 years12
Vilniaus apskritisLT#111 years10

München held the top position among 401 German districts for every year in the dataset — the most statistically robust case of persistent leadership in the sample.

Permanent seconds

RegionCountryRankDurationBehind
AnkaraTR#211 years&İstanbul
Västra Götalands länSE#211 yearsStockholm
Bălţi MunicipalityMD#211 yearsChişinău
Dnipropetrovsk OblastUA#2–310 yearsKyiv

Permanent bottom

RegionCountryRankDurationTotal regions
Karlovarský krajCZ#14 (last)11 years14
TicinoCH#26 (last)11 years26
NógrádHU#20 (last)11 years20
Zaječarska oblastRS#25 (last)8 years25
What stability means

Persistent stability is not inherently positive or negative. München at #1 and Karlovarský kraj at #14 both show zero rank variance, but they represent opposite ends of the distribution.

Regional hierarchies within countries are largely fixed over the medium term. When they do change (as in Turkey’s earthquake or Greece’s employment recovery), it is typically driven by an external shock or a structural transition — not by gradual drift.

A region moving 5 positions in a country with 80 regions is proportionally small (6%) and may reflect normal fluctuation. A region moving 20 positions signals something structural.


5. Methodology

Composite construction

The composite index aggregates official statistics into a single comparable score per region per year. The process is transparent and reproducible:

  1. Collect. Raw indicators from Eurostat (NUTS-3 and NUTS-2 levels) and national statistical offices. Only official published data is used. No estimates, projections, or proprietary data.
  2. Normalize. Each metric is min-max normalized within its country and year: (value − min) / (max − min). This produces a 0–1 score where 0 = lowest in the country that year, 1 = highest.
  3. Assign axes. Metrics are grouped into two axes for EU countries:
    • Economic (weight 60%): GDP per capita, employment rate, unemployment rate, GVA by sector, long-term unemployment
    • Demographic (weight 40%): population, growth rate, net migration, birth/death rates, median age, life expectancy, dependency ratios
  4. Aggregate. Each axis score is the average of its normalized metrics. The composite = 0.6 × economic + 0.4 × demographic, scaled to 0–100.

What the composite does not do

Limitations

Full methodology: sky-mind.com/methodology


6. Data Appendix

Full rank tables by country available at sky-mind.com/map.html. Raw data available for download at sky-mind.com/data. The lineage of any figure in this review — official source, dataset, original value, load date — is one click away at sky-mind.com/provenance; the underlying Eurostat tables are at ec.europa.eu/eurostat.

Coverage summary

CountryRegionsYearsEconDemoTotal
DE4012014–2024211444
UK1822023134
IT1072014–2024231447
TR812014–2024151438
PL732014–2024201445
ES592014–2024201445
GR522014–2024211445
BE442014–2024211446
RO422014–2024231448
NL402014–2024211444
AT352014–2024201445
MD352014–2024134
BG282014–2024231448
CH262014–2024111436
UA252004–2021134
PT252014–2024231448
RS252017–2024201344
HR212014–2024231448
SE212014–2024231447
HU202014–2024231447
FI192014–2024231448
CZ142014–2024231448
NO122014–2024211445
SI122014–2024211446
AL122014–202321422
DK112014–2024231448
LT102014–2024231448
MK82014–2024201441
SK82014–2024231448
IE82014–2024211446
LV62021231448
EE52014–2024231447

32 countries, 1,948 regions. Metric counts reflect unique indicators in the raw dataset; not all metrics are available for all regions or years. UA data suspended after 2021 (war). UK, MD have limited metric coverage (national sources).


SkyMind Regional Review is published by SkyMind Analytics. This review reflects the latest official statistics available at the time of publication. The composite index is a descriptive measure of relative regional position within each country. It is not a prediction, recommendation, or assessment of quality of life.

Methodology · Data · info@sky-mind.com

© 2026 SkyMind Analytics. Data sources: Eurostat, TurkStat, CZSO, BFS, INSTAT, and national statistical offices.