Empty homes in Luxembourg: the only complete count is from 2021 (15,337 dwellings) and it counts missing registrations

How many homes are empty in Luxembourg?

We found no official count that answers that, because the figures in circulation are two data sets, two State statements and a mayor's statement about one city, and each counts something different.

Only the first two rows come from data collected on dwellings, and they cannot be averaged, added or read as a trend, because their bases differ. The census row covers all conventional dwellings and is a register check. The survey covers only the dwellings declared by the people who answered, and only 20.8% of the 27,889 owners contacted answered. The 6% is a rate applied to a stock figure, not a count: 6% of about 244,200 is 14,652, which L'essentiel reports as 14,600, and the Housing Ministry described about 6% as the share always empty because of market movement, about 15,000 dwellings (Paperjam, 1 April 2021). Its closeness to the census's 5.8% is therefore not a cross-check, and we keep it out of any trend. The second chart on this page shows the three shares (5.8%, 2.3% and 6.0%) side by side, each on its own base; its labels are in English. The Housing Minister, Claude Meisch, answered "Around 10,000" when Paperjam asked him (interview published 3 October 2025) how many vacant homes he estimates there are; no method or base was given, so we do not use it as data either. On 28 July 2026 the mayor of Esch-sur-Alzette, Christian Weis, said on RTL that about 8% of the city's apartments are empty and that he wants to introduce a vacancy tax from autumn 2026, of up to 10,000 euros per unoccupied apartment depending on size and duration (L'essentiel, 29 July 2026). No method was given, so we do not use the 8% as data either, and it cannot be compared with the register's 6.9% for Esch (the definition and date of the 8% are not stated). We found no report of a council vote on that tax as of 9 October 2026. The 2021 table is the only one of these figures that covers all conventional dwellings and has a breakdown by commune.

L'essentiel reported the State's 6% on 14 October 2021, 25 days before the census date, and quoted the Housing Ministry: "we have no reliable data today". The paper also reported, citing the Observatoire de l'Habitat, vacancy of 7.7% in Luxembourg City in 2010 and 7.1% in Esch-sur-Alzette in 2012. It did not state how those were measured, so we do not use them as a trend line.

Which communes have the most homes with no registered resident?

The census table splits the 15,337 by commune. Luxembourg City holds 6,189 of them, 40% of the national count, with 23% of the country's dwellings (61,920 of 263,671).

STATEC census table, 2021, communes with at least 5,000 dwellings; chathome.lu calculation of the shares. The first chart on this page repeats these shares; its labels are in English.

Only 10 of the 102 communes are above the national 5.8%. The highest shares are in small places: Weiswampach (158 of 1,133, 13.9%) and Echternach (292 of 2,524, 11.6%). At the other end, Waldbredimus has 0 of 444. In a commune of 444 dwellings, every ten dwellings move the share by more than two points, so the small-commune ranking is too unstable to compare. For prices and live listings in the capital, see the Luxembourg commune page and its price page.

Worked example: Esch-sur-Alzette. The table lists 15,906 dwellings, of which 1,102 had no registered resident: 1,102 ÷ 15,906 = 6.9%, against 5.8% nationally. Because it is a count of registrations on one day in 2021, it does not say that 1,102 homes were empty for a year, that any owner meant to leave one empty, or that the figure is the same today. In July 2026 the city's mayor put the share of empty apartments at about 8%, a statement with no method or definition given; we do not read the gap with 6.9% as a change.

What is my commune's share? All 102 communes in 2021

Find your commune below. Each share is STATEC's count of dwellings with no registered resident divided by all dwellings in the commune, sorted from the highest share; the rows add up to the national 15,337 of 263,671 (5.8%). These are the 102 communes of the 2021 census; L'essentiel counts 100 communes in 2026. The ten communes above 5.8% are Weiswampach, Echternach, Luxembourg, Hesperange, Boulaide, Esch-sur-Alzette, Diekirch, Esch-sur-Sûre, Mamer and Strassen. Because the shares of small communes move a lot with a few dwellings, read them as registration counts, not as a ranking.

STATEC census table DF_B1707, 2021; chathome.lu calculation of the shares, rounded to one decimal.

Does "no registered resident" mean "empty"?

No, and the method as Paperjam describes it shows why. Paperjam (26 August 2026) reports that the 2021 figure was not an estimate and not a question to residents: the census questionnaire had no heading on unoccupied dwellings. STATEC matched its building-and-dwelling register against the national register of natural persons, and a dwelling with no registered resident counted as unoccupied. The European definition STATEC follows also puts second and seasonal homes, and homes under renovation or newly built with no one registered, in the same group. In the other direction, a home in poor condition or under works still counts as occupied if someone keeps their official residence there.

The table itself shows the pattern, as of census day (8 November 2021). Of the 15,337 dwellings, 13,186 (86%) are in buildings with three or more dwellings, which hold half of all dwellings (131,944 of 263,671); there, 10.0% had no registered resident. In two-dwelling buildings the share is 14.9% (1,915 of 12,880); in one-dwelling buildings it is 0.2% (236 of 118,847). By construction period the shares for dwellings built before 2011 run from 4.1% (1981 to 2000, 2,183 of 53,011) to 6.4% (1919 to 1945, 1,747 of 27,329); 2011 to 2015 is 6.7% (1,315 of 19,714) and 2016 or later is 10.4% (2,755 of 26,540). A further 5,329 dwellings have no recorded period, 780 of them unoccupied. Dwellings built in 2016 or later are 10.1% of the stock (26,540 of 263,671) but 18% of the count (2,755 of 15,337). The table does not say why, and we have not tested an explanation. Two sources describe registration practice. Paperjam reports that the planned dwelling register would let communes register residents in a specific dwelling rather than at an address shared by several households. In 2021 the president of the communes' association Syvicol told L'essentiel that communes know from their population register how many people are housed but mostly not how many dwellings they are spread across. That is consistent with a concentration in buildings of three or more dwellings, but the table cannot show how much of the 13,186 it explains, and we do not call it a cause. The national figure leans on larger and more recent buildings.

The building mix also explains part of the gap between communes. Luxembourg City has 81% of its dwellings in buildings of three or more dwellings (49,902 of 61,920), Pétange 48% (3,649 of 7,639). Inside those larger buildings only, the large communes sit closer together: 11.3% in Luxembourg City (5,637 of 49,902), 9.0% in Esch-sur-Alzette (1,008 of 11,154), 9.1% in Sanem (251 of 2,761), 8.3% in Differdange (549 of 6,611), 8.1% in Pétange (295 of 3,649) and 7.9% in Dudelange (323 of 4,086), with Hesperange highest at 12.5% (458 of 3,651).

What does the 2025 survey add, and who is missing from it?

In a survey run for the Housing Ministry by LISER, letters went out on 22 October 2025, with a reminder on 11 November 2025, and 27,889 individuals who, in the land registers, own several properties were contacted. 5,799 answered (20.8%) and declared 16,593 dwellings: 9,401 let, 5,510 main residences, 566 occupied free of charge, 384 second residences, 353 "other" and 379 unoccupied. That is 2.3% of the dwellings declared (the report rounds to 2%). Two features limit it:

• Declared and cadastral counts differ. The respondents declared 4,048 more dwellings (32.3%) than the 12,545 the cadastre attributes to them. The report says this confirms a gap between a property's cadastral category and its actual use.
• Legal entities were not asked. About 5,500 legal entities, public and private, hold roughly 9,300 apartments and 8,800 houses, about 18,100 dwellings, and received no questionnaire. A first classification puts about 30% of that stock in public hands (funds, communes, the State). Because they were not surveyed, the report has no figure for how many of these dwellings are unoccupied.

The two counts are also defined differently: the census count includes second homes, while the survey reports 384 second residences separately from its 379 unoccupied dwellings. Response fell from 21.2% among owners of two dwellings or fewer to 14.4% among owners of six or more.

Paperjam notes that the survey adds something the census lacked, a separate answer for second residences and unoccupied dwellings, but that the 2.3% cannot be applied to the whole housing stock. We agree: it is what some individual multi-owners say about their own homes.

Why is there no newer count?

STATEC calculates unoccupied dwellings only at a census, according to Paperjam (26 August 2026), which reports that the institute is unlikely to measure again before the 2031 census. The institute told the paper it had never commented on or analysed the 2021 result. Two bills that would change this were in committee when we checked the Chamber's dossier pages on 9 October 2026:

• Bill 8086, filed on 19 October 2022, would create a national register of buildings and dwellings giving every unit its own identifier. The Chamber's dossier shows "En commission" and no published step since 24 November 2023, when it was referred to the Commission du Logement et de l'Aménagement du territoire; no amendment is listed. Paperjam reports the government's own estimate of up to three years to build the register once the law applies, and says that without the bill that timetable cannot start. In a joint written answer of 30 September 2025 to LSAP parliamentary question n°2806, Interior Minister Léon Gloden and Housing Minister Claude Meisch said that bill 8086 was being amended, with talks planned with Syvicol and the City of Luxembourg before the amendments are tabled, that the register is a precondition for introducing a tax on non-occupied dwellings, and that a reliable overview of occupied and vacant dwellings has been missing to date; the dossier shows no step after 24 November 2023, as of 9 October 2026.
• Bill 8082B, split from the property-tax reform in July 2025, would tax dwellings with no habitual resident for six consecutive months, subject to exceptions (Paperjam). The Chamber's dossier shows it in committee, last updated 3 July 2025, with one parliamentary amendment of that date. Paperjam reports the government expected at least 12 more months after the register for the tax.

Some communes already tax unoccupied buildings: L'essentiel reported on 14 October 2021 that five of the 102 had put the tax in place. On 29 July 2026 it reported that eight communes have one (Esch-sur-Alzette, Beckerich, Diekirch, Echternach, Esch-sur-Sûre, Redange-sur-Attert, Roeser and Winseler) and that Esch-sur-Alzette's tax, in place since 2012, has never been levied. The two counts come from two reports (five of 102 communes in 2021, eight of 100 in 2026) and should not be read as a trend. The trade-off is live: Esch-sur-Alzette's mayor wants a tax from autumn 2026, while bill 8086 has had no published step since 24 November 2023. The guichet.public.lu page (last updated 2 November 2015) says a building counts as unoccupied when no one is listed in the population registers as living there, or using it as a secondary residence, for 18 consecutive months, and that communes may waive the tax in full or in part. That is a legal test for a communal tax, not a measure of the housing shortage, because it asks only about registration. On 10 July 2026 two LSAP deputies also asked the ministers of Finance and Housing, in question 4434, how many State-owned dwellings are vacant, by commune and by length of vacancy.

What is the strongest case for the 2021 figure, and what would change our reading?

The case for it is fair. It is the only count that covers every conventional dwelling in the country, uses one definition everywhere, and comes from administrative registers rather than from answers people chose to give. For comparing large communes with each other on that one definition, it is usable, and a capital at 10.0% against 4.2% to 7.8% in the other large communes is a real difference in registrations. "Vacant" also does not mean "unused": the group can include second homes, renovations and new builds waiting for their first resident.

Our position is narrower. No figure in circulation says how many homes are empty, for how long, or whether they could be let, and a reader or a commune should not set policy or a story on one until the date, definition and universe are checked. This reading would be wrong if STATEC or the Ministry publishes a newer count with a documented method, or if a documented, reproducible method is shown for the State's 6% or the minister's 'around 10,000'. It would also be wrong if the 2021 breakdown by building type were shown to separate genuinely empty buildings from registration gaps.

The trade-off for a commune. Waiting for the register means waiting for a bill that the ministers said in September 2025 was being amended but that has had no published step since 24 November 2023, with STATEC unlikely to measure again before the 2031 census. Acting now means using a registration-based test that communes have been allowed to apply since October 2008 and that, by one L'essentiel count (29 July 2026), only eight communes have adopted. Our reading: a commune that taxes now should say which test it applies and what it exempts, and not cite the 2021 share as the size of the problem; a commune that waits should not describe the wait as measurement.

What can a reader do with an "empty homes" figure?

Ask three questions before using any figure: what date, what counts as "empty", and which dwellings are in the base. Then compare your commune's 2021 share with the national 5.8% and with the live market on the buy page, and read how our own asking-price figures are built in the price methodology. To check your own commune, find it in the table of all 102 communes above; the source is the Lustat table.