ImmigrationDBVerified immigration data

Where foreign populations grew, unemployment fell

By · Published · Updated

Question: in cities where the foreign population grew, did the job market get better or worse?

Answer: better — clearly, inside every country large enough to measure, and invisibly when the countries are pooled. Across 359 cities the pooled coefficient between foreign-population change and unemployment change is 0.12, which reads as no relationship at all. Split by country it reverses and strengthens: −0.41 in Germany across 108 cities, −0.34 in Italy across 74, −0.15 in France across 45, −0.10 in Spain across 79.

Negative is the favourable direction here: the cities that gained the most foreign residents are the ones whose unemployment rate moved down the most. What this cannot tell you is which came first. That question is beyond this data, and the article does not answer it in either direction.

What we measured

One row per city, for cities where both series have a usable run over an overlapping period. foreign_population_total is the published stock of foreign residents, measured as the percentage change between the first and last observation. unemployment_total_percent is the published unemployment rate, measured as the change in percentage points. After the validation in the method note below, 359 cities across 15 countries qualify. The median row has nine overlapping annual observations.

Two cautions belong up front. Foreign population is a stock, not a flow — it counts people recorded as resident on one date, so naturalisation removes someone without a move, and a rise is never "arrivals". And unemployment rates are national constructions: each statistical office defines the labour force, the reference week and the registration rule its own way. So this article compares the direction of change within a city, and cities within one country against each other. It never ranks unemployment levels across borders.

The sample is concentrated. Germany contributes 108 cities, Spain 79, Italy 74 and France 45 — 306 of 359 rows. Bulgaria adds 17, Sweden 10, and the remaining nine countries contribute fewer than five each.

Both measures are endpoint comparisons. A city that improved early and stalled looks identical to one that deteriorated and then recovered. That limitation is what stops the headline question — follow or lead — from having an answer here.

What the data shows

The most common combination is more foreign residents and lower unemployment: 231 cities, 64.3%. A further 103 cities (28.7%) ended with both higher. Foreign population fell in only 25 cities, 20 of them alongside falling unemployment.

Counting quadrants only records where each city ended. The coefficient is what shows whether the two changes are related — and it depends entirely on whether the countries are pooled or separated.

Germany is the sharpest case and it needs care, because its two numbers point in apparently opposite directions. Most German cities in the sample — 74 of 108 — finished with unemployment higher than at the start, since their windows run through 2020–2022. Yet the German coefficient is −0.41. Both are true: German unemployment generally rose over these windows, and the cities that gained the most foreign residents are the ones where it rose least or fell.

Italy is −0.34 across 74 cities, with 58 of them in the foreign-up, unemployment-down quadrant. France is −0.15 across 45, Spain −0.10 across 79. The direction is consistent; only the strength varies.

The cities with the largest foreign-population growth among those where unemployment fell are mostly eastern German. Gera grew its foreign population 819.7% between 2011 and 2022 while unemployment fell from 11.2% to 7.6%. Görlitz grew 458.6% with 16.3% to 9.2%; Plauen 410.3% with 11.6% to 6.9%; Erfurt 360.0% with 8.4% to 6.3%; Magdeburg 325.9% with 9.5% to 8.8%. Vilnius grew 561.1% between 2011 and 2024 while unemployment fell from 13.1% to 4.7%, Trondheim 349.8% with 3.0% to 1.8%, and Burgas 319.6% with 5.7% to 2.1%.

Those percentages are large because the starting counts were small. A city going from 1,479 to 13,603 foreign residents has multiplied its foreign population ninefold and still has a modest one in relative terms. The rate and the base must be read together.

Where it breaks down

The pooled 0.12 is the number to distrust. It is computed across 15 national unemployment definitions, different windows and very different levels, and it lands near zero because the countries partly cancel one another. Every large country inside it is negative. Quoting the pooled figure as "no relationship between migration and jobs" would be the opposite of what the country data shows — this is the same structure as pooling any set of groups whose internal patterns disagree with the aggregate.

The second thing that breaks is the direction of the arrow. A −0.41 coefficient in Germany is consistent with at least three stories the data cannot separate: people moved to cities where work was available; cities that received people saw their labour markets tighten; or a third factor — an industrial base, a university, a housing market — drove both. Nothing in this packet distinguishes them, and no wage, vacancy or occupation variable is present.

The 103 cities where both measures rose are the useful counterweight. They are not an exception to a rule; they are more than a quarter of the sample, and they show that a growing foreign-resident population is entirely compatible with a rising unemployment rate over the same window.

What this means if you are moving

Do not use a city's foreign-population trend as a proxy for its job market, or the reverse. Even at −0.41, knowing one leaves most of the variation in the other unexplained, and for a single city the coefficient predicts nothing.

Read both series for the city and the years that matter to you, under that country's own definitions. A falling unemployment rate under Germany's registration-based measure is not the same statement as a falling rate under Spain's survey-based one, and neither describes your occupation.

Take the base into account before being impressed by a growth figure. Several cities near the top of the table multiplied a foreign population of one or two thousand. That is a real change in the city and a small absolute community — two facts that matter differently depending on what you are looking for.

And treat the country you are considering as its own case. A mover looking at Germany meets a sample where most cities' unemployment rose over the window; a mover looking at Spain meets one where it fell almost everywhere. Those are different periods and different labour markets, not competing theories.

Limits

The analysis cannot identify timing. Each series is reduced to a first-to-last change, so no turning point can be dated, no lag tested and no sequence established. The title's question has no directional answer from this packet, and none is offered.

Foreign population is a stock under national definitions that differ, and unemployment follows national labour-force conventions that also differ. Within-city direction is comparable; levels across countries are not.

Growth from a small base inflates percentages. The largest changes in Table 2 start from counts in the low thousands, and a few hundred people move those figures a long way.

Coverage is concentrated and uneven: four countries supply 306 of 359 cities, and start years, end years and observation counts vary between rows.

Finally, the packet holds no wages, vacancies, occupations, housing or reasons for moving. It supports a description of how two official series moved together, not an explanation of why any city changed.

Method

The packet supplied 444 cities. Eighty-five were removed, and the reason is a result in itself.

Eighty of them draw on a foreign-population series that fails a continuity test applied to every trend in this research series: the publisher changes at the final observation with a jump of more than 50%, or a single year stands more than twice above both its neighbours. Every Dutch city in the sample fails — their series end on a national migration-background count roughly four times the preceding citizenship count — along with 21 French cities and small numbers in Italy, Finland, Germany, Spain, Norway and Belgium. The submitted Table 2 was led by five Dutch cities showing 889% to 1,136% growth; those figures measure a change of definition.

Five further rows, all Slovak, report zero overlapping observations with a last year of 2009 against a first year of 2012 — impossible under the inclusion rule and removed.

On the remaining 359 cities, foreign-population change is the percentage change between first and last observation; unemployment change is the last rate minus the first, in percentage points. Spearman's rank correlation is the primary coefficient because the two measures are on different scales and the percentage changes include very large values. Each city is weighted equally; no population weighting is applied.

A country receives its own coefficient only with at least 20 qualifying cities, which leaves Germany (108), Spain (79), Italy (74) and France (45). Quadrant counts are 231, 103, 20 and 5.

Sources

Frequently asked questions

Can these data show whether migration follows jobs?

No. Each city is reduced to the change between its first and last annual observation, so the analysis cannot date a turning point, test a lag, or say which trend moved first. It can show that the two changes are related within countries, and it cannot say which one caused the other.

Do cities that gain foreign residents have worse unemployment?

The opposite, inside every country large enough to measure. Spearman's correlation between foreign-population change and unemployment change is −0.41 in Germany across 108 cities, −0.34 in Italy across 74, −0.15 in France across 45 and −0.10 in Spain across 79. Negative means the cities with the largest foreign-population growth are the ones where unemployment moved most favourably.

Why is the pooled number so different from the country numbers?

Pooling 15 national systems with different unemployment definitions and different periods produces 0.12, which reads as no relationship. Every large country inside it is negative. The pooled figure describes the mix of countries in the sample, not any city in it.

Which cities combine the largest foreign-population growth with falling unemployment?

Gera, up 819.7% between 2011 and 2022 with unemployment falling from 11.2% to 7.6%, then Vilnius at 561.1% with 13.1% to 4.7%, and Görlitz, Plauen, Erfurt and Magdeburg. Most are eastern German cities growing from a very small foreign-resident base.

Why were 85 cities excluded?

Eighty of them draw on a foreign-population series that changes definition or contains a corrupt year — every Dutch city in the sample is affected. Five Slovak rows report zero overlapping observations with a last year earlier than their first, which is impossible under the inclusion rule.

More research

More foreign residents, more crime? The answer changes at every border

Does a larger foreign population mean more recorded crime? It depends on the country: 0.06 in Germany and the US, 0.50 in Spain and Belgium, 0.59 in the Netherlands.

Clean air barely shows up in the rent

Do you pay more rent for cleaner air? Across 392 German housing markets the link is weak — Spearman −0.12 — and 31 districts sit below the median on both.