Distribution of Luxembourg's cross-border workers by commune: the 783 communes concerned

Luxembourg covers 2,586 km², but the population that sustains its housing market lives across 783 communes in three countries. The Luxembourg social-security register makes that distribution measurable: it records, commune by commune, where the people who work in Luxembourg without living there reside.

This analysis is strictly demographic. It describes where people live and how far they are from work; it contains no cross-border price comparison, as no usable, licensed transaction data exists for France, Belgium and Germany.

The IGSS (Inspection générale de la sécurité sociale) publishes, twice a year, the number of employed people affiliated to Luxembourg social security, broken down by commune of residence in France, Belgium and Germany. At the 2026-03-31 reference date those files attribute 228,990 cross-border workers to a named commune, spread across 1,367 communes. We applied a published inclusion rule to that list, and the result is the map below.

783 communes, 218,690 cross-border workers: what the register shows

Of the 1,367 communes housing at least one Luxembourg-affiliated worker, 783 fall inside the frontalier belt. They hold 218,690 people — roughly 95.5 % of all cross-border workers the register attributes to a commune.

So a little over half the candidate communes capture more than nineteen in twenty cross-border workers. That is not a coincidence; it follows directly from how this population is distributed in space: heavily clustered near the border, heavily diluted beyond it.

Why the belt is not a circle

A radius drawn around Luxembourg City would produce a false map, because cross-border density does not fall off smoothly with distance. It falls in steps, and the steps differ on all three sides of the border.

The 15–30 km band alone holds 39.6 % of covered workers across 155 communes. The next band, 30–45 km, contains nearly twice as many communes — 296 — but only 22.6 % of the workers. That inversion is the central fact of this geography: moving outward adds a great many communes and very few people.

For example, for anyone house-hunting: past 45 km, each additional commune carries fewer than a hundred people on average. The density of the cross-border community — and therefore the depth of the rental market serving it — collapses well before the 100 km limit.

France, Belgium, Germany: three geographies, three logics

The three countries do not look remotely alike, and the reason is administrative as much as demographic.

France is concentrated. 343 of the 407 French communes included qualify on proximity, and they carry 110,220 workers — the bulk of the French contingent. The Thionville–Longwy–Villerupt corridor is a continuous built-up band oriented towards Luxembourg: Thionville alone holds 10,110 cross-border workers at 20 km.

Belgium is the opposite. Only 15 Belgian communes qualify on proximity, yet they carry 28,680 people, because Belgian communes are large: Arlon (8,670 workers) and Aubange absorb what in Germany would be split across twenty villages. Belgium is also the only country from which the distance cap removes meaningful volume — 22 communes and 2,310 workers — which is why its coverage is the lowest at 91.1 %.

Germany is fragmented. It takes 319 communes to cover 48,890 workers, because a Rhineland-Palatinate Ortsgemeinde can be a tenth the size of a Walloon commune. Apart from Trier, Stadt (8,680) and a few towns such as Perl (2,180), the German contingent is a scatter of very small communes: 159 of them qualify on the volume test alone, for 10,070 people in total.

The communes that carry the most weight

The top 10 communes hold 57,660 cross-border workers, 26.4 % of the covered total; the top 50 hold 53.5 %. Only three towns — Thionville, Trier, Stadt, Arlon — exceed 8,000 people each, and they sit in three different countries.

One case is worth flagging because it illustrates the rule: Metz, at 48 km, does not qualify on proximity but on volume (6,330 workers). The same holds for Bastogne (3,720 at 56 km) and Saint-Vith (1,770 at 74 km). These communes are far away, but the number of people leaving them each morning is too large to ignore.

The long tail: 455 communes with fewer than a hundred workers

This is the least intuitive fact in the dataset. Of the 783 communes in the belt, 455 — 58 % — hold fewer than a hundred cross-border workers. The median commune in the belt is Püttlingen, Stadt, with 70 workers at 62 km.

For example: Thionville concentrates 10,110 cross-border workers at 20 km, while the belt's median commune, Püttlingen, Stadt, holds 70 at 62 km — a ratio of more than one hundred and forty to one. Two communes in the same belt, two incomparable market depths: a mass cross-border population in the first, a few dozen households in the second.

A map of the cross-border population is therefore a handful of dense poles laid over a very long carpet of villages. Any account that keeps only the big towns misses 73.6 % of the phenomenon; any account treating all 783 communes as equivalent is just as wrong.

Why we do not publish the institutional Grande Région

An official cross-border perimeter already exists: the Grande Région, a cooperation body whose geographic manifest we maintain lists 5,034 communes (2,321 in France, 2,352 in Germany, 261 in Belgium, 100 in Luxembourg). We do not use it as a market perimeter, and the reason fits on one line of data.

The Grande Région's Belgian partner is the entire Walloon Region, so Charleroi belongs to it. But Charleroi has 70 residents affiliated to Luxembourg social security, 148 km away. Liège has 360 at 120 km, Namur 170 at 131 km. Treaty membership is not evidence that anyone commutes.

Publishing a housing market in those cities because a cooperation agreement names them would mean inventing demand that does not exist. We prefer a measured rule, because it turns "why is my commune not covered?" into a number. Of the 5,034 institutional communes, 783 pass the test.

Method, parameters and limits

The rule has two legs; meeting either one is enough.

• Volume and range: at least the national floor of Luxembourg-affiliated residents (40 in France, 50 in Belgium, 20 in Germany) and less than 100 km away.
• Proximity: within 45 km, any cross-border presence at all qualifies.

Distances are straight-line, measured to the nearest of five Luxembourg employment centres: Luxembourg City, Kirchberg, Cloche d'Or, Belval and Findel.

The three source files are public and freely downloadable from IGSS under a CC0 1.0 licence: emploi-total-communeres-f.xlsx (France), emploi-total-communeres-b.xlsx (Belgium) and emploi-total-communeres-d.xlsx (Germany), at https://igss.gouvernement.lu/dam-assets/publications/statistiques/emploi/emploi-total-communeres-f.xlsx. Anyone can therefore reproduce this calculation from the same files and the parameters published above, and arrive at the same 783 communes. Because counts are rounded to the nearest ten at source, the 95.5 % coverage rate inherits that uncertainty and should be read as an order of magnitude, not an exact value.

The floors differ because commune sizes differ by an order of magnitude between countries. They are derived, not hand-picked: each is the smallest step on the IGSS rounding grid that still retains roughly 95 % of that country's commune-attributed cross-border workers. IGSS rounds every published cell to the nearest ten for disclosure control — which is why the German floor is 20 and not 10: a commune reporting 10 could really hold 5.

The 100 km cap exists for one specific reason, and it is this source's main limitation: IGSS measures a registered address, not a commute. Someone who kept an address in Hainaut or the Brussels orbit appears in the file exactly like a daily commuter. Beyond 100 km we treat the registration as an administrative address rather than evidence of commuting.

Two further limits deserve stating. First, IGSS publishes no geographic codes — no INSEE, no REFNIS, no AGS — so the join to the commune reference runs on normalised names; a commune merger can therefore drop a row silently, which we guard with a minimum join rate. Second, people the register attributes to no commune are discarded and never redistributed pro rata, because spreading them would invent demand where there is none. Commune names appear as IGSS publishes them, so every row stays traceable to the source file.

This map describes demand, not prices. We publish no cross-border price comparison: we hold no usable, licensed transaction data for France, Belgium and Germany, and we would rather say nothing than quote a figure we could not source. On the Luxembourg side our pricing method is published in full — see the price methodology and prices by commune. You can also browse properties for sale in Luxembourg.