Marius Comper

Housing · 2.946 localities · 1992–2025

Romania Keeps Building. But For Whom?

Romania's national statistics office counts, every year, how many dwellings exist in every locality. Since 1992, the country's housing stock has grown 29,5%, while the domiciled population fell 8,3%. That is not a coincidence: in 65,2% of the 2.946 localities with complete data, the housing stock grew exactly where the registered population left.

65,2%of localities have more housing and fewer domiciled residents than in 1992 — the "paradox" quadrant
+29,5%growth in the national housing stock, from 7.681.014 to 9.944.989 dwellings, 1992 to 2025
−8,3%fall in domiciled population in the same localities, from 23.143.860 to 21.227.988
Băișoara(Cluj county) holds the record: +124% housing, −30% domiciled population since 1992

The explorer

Your locality, on the map

Search for your locality or click it on the map. The card on the right shows how many dwellings and how many domiciled residents it had in 1992 and now, and which of the four quadrants it falls into. The buttons above switch what the map is coloured by.

Loading the map…

The quadrant compares the change in housing stock with the change in domiciled population, 1992→2025, for each locality. "No full data" means localities created after 1992, or with a gap in one of the two series.

The paradox, in plain sight

Every dot is a locality

The horizontal axis shows how much the housing stock grew or shrank since 1992; the vertical axis shows how much the domiciled population grew or shrank, over the same period. A dot in the bottom-right corner — more housing, fewer people — is a dot in the paradox quadrant. There are 1.920 such dots, more than in any other quadrant.

-100%-50%0%50%100%150%200%-100%-80%-60%-40%-20%0%20%40%60%expansionparadoxcontractionpressureChange in housing stock, 1992→ 2025 (%)Change in domiciled population, 1992→ 2025 (%)
2.946 localities, percentage change in housing stock (horizontal) against percentage change in domiciled population (vertical), 1992→2025. Axes are capped at +200% (housing) and −100%/+60% (population) for readability; a few localities with extreme housing growth (the Cluj, Ilfov, and Timiș commuter belts) fall outside this frame on the horizontal axis, but remain in the expansion quadrant, not paradox.
1.920paradox — more housing, fewer people
614expansion — both grew
402contraction — both shrank
10pressure — only population grew

The extreme cases

Three rankings, three different mechanisms

The localities with the largest paradox gap are not the poorest villages, but often mountain or weekend resorts, where second homes get built into a shrinking resident population. The localities with the largest population loss are small, isolated villages where the housing stock shrank too — abandoned houses, not new ones. The localities with the largest housing growth are almost all commuter-belt communes around the big cities, where both housing and people arrive together.

The biggest paradox

  • 1BăișoaraCluj county+124%housing, -30% population
  • 2BelceștiIași county+134%housing, -17% population
  • 3FundataBrașov county+86%housing, -54% population
  • 4DâmbovicioaraArgeș county+100%housing, -36% population
  • 5TarcăuNeamț county+125%housing, -7% population
  • 6BelciugateleCălărași county+104%housing, -16% population
  • 7SpermezeuBistrița-Năsăud county+96%housing, -23% population
  • 8ErbiceniIași county+115%housing, -2% population
  • 9Moldova NouăCaraș-Severin county+87%housing, -28% population
  • 10VoitegTimiș county+109%housing, -2% population
  • 11BuciumeniGalați county+91%housing, -19% population
  • 12SâmbureștiOlt county+68%housing, -40% population
  • 13AlbeniGorj county+101%housing, -7% population
  • 14FarcașaNeamț county+98%housing, -10% population
  • 15FârdeaTimiș county+78%housing, -29% population

The biggest population loss

  • 1MalnașCovasna county-80%population, -72% housing
  • 2Slobozia MândraTeleorman county-74%population, -50% housing
  • 3Optași-MăguraOlt county-74%population, -46% housing
  • 4BărbulețuDâmbovița county-74%population, -54% housing
  • 5BaldovineștiOlt county-74%population, -61% housing
  • 6Ciceu-GiurgeștiBistrița-Năsăud county-73%population, -54% housing
  • 7TanacuVaslui county-72%population, -42% housing
  • 8RăduleștiIalomița county-72%population, -48% housing
  • 9ArmășeștiIalomița county-71%population, -32% housing
  • 10RusSălaj county-71%population, -55% housing
  • 11CiuceaCluj county-70%population, -51% housing
  • 12VădastraOlt county-70%population, -52% housing
  • 13BârseștiVrancea county-70%population, -53% housing
  • 14Albeștii de MuscelArgeș county-69%population, -47% housing
  • 15Nicolae TitulescuOlt county-68%population, -48% housing

For contrast: the biggest housing growth

  • 1FloreștiCluj county+1473%housing, +893% population
  • 2ChiajnaIlfov county+1324%housing, +431% population
  • 3GirocTimiș county+1260%housing, +586% population
  • 4DumbrăvițaTimiș county+1238%housing, +731% population
  • 5BragadiruIlfov county+944%housing, +486% population
  • 6MiroslavaIași county+940%housing, +423% population
  • 7Popești-LeordeniIlfov county+704%housing, +293% population
  • 8ȘelimbărSibiu county+653%housing, +303% population
  • 9Moșnița NouăTimiș county+643%housing, +400% population
  • 10BerceniIlfov county+626%housing, +193% population
  • 11SânpetruBrașov county+609%housing, +337% population
  • 12CorbeancaIlfov county+578%housing, +213% population
  • 13TunariIlfov county+574%housing, +171% population
  • 14Ștefăneștii de JosIlfov county+504%housing, +205% population
  • 15DobroeștiIlfov county+470%housing, +143% population

These localities are not in the paradox quadrant — their population grew as fast as, or faster than, their housing stock. They are included for contrast: the paradox is not about "where a lot gets built," but "where building happens without people following."

Why this happens

Four plausible explanations, no direct proof

The TEMPO data shows what happened, not why. It cannot separate a new house lived in year-round from a second home used two weeks a year, and it cannot say who built it or with what money. Four mechanisms are nonetheless consistent with the pattern observed, without the data here being able to confirm or rule them out directly:

  • Second homes and weekend tourism. Several of the paradox records — Băișoara, Fundata, Dâmbovicioara — are mountain or border-tourism localities, where part of the new housing stock serves visitors or owners domiciled elsewhere.
  • Remittances. Romania has one of the largest diasporas relative to population in the EU; a house built or extended in a home village with money sent from abroad does not automatically bring back a registered domicile.
  • Split inheritance. A single household inherited by several siblings can, over time, become several separately registered dwellings, with no change in population at all.
  • Domicile does not follow the move. Someone who actually moves from a village to a city can keep their legal domicile at the old address for years; a village's "real" population may be even smaller than POP107D shows, not larger.

Method

Where the numbers come from, and what they don't say

The data come from the TEMPO-Online database of Romania's National Institute of Statistics: matrix LOC101B (dwellings existing at year end, all forms of ownership, by county and locality, 1990–2025) and matrix POP107D (population by legal domicile on 1 January, by county and locality, 1992–2026, reused from earlier pulls for the "When Did Modernity Reach You?" and "How Old Is Your Locality?" projects). Both were accessed directly through the TEMPO-Online API (`tempo-ins/pivot`) on 9 September 2026.

  • The common window. The comparison uses 1992 (the first year with locality-level population data) and 2025 (the last complete year in both series). Of Romania's 3,181 localities, 2,946 have valid, non-null data in both matrices in both years; the rest were created after 1992 or have a gap in one series, and are excluded rather than estimated.
  • What LOC101B counts. The stock of existing dwellings, regardless of ownership, at the end of each year. It says nothing about occupancy: a counted dwelling may be lived in year-round, part of the year, or not at all. For the question "how many houses stand empty," see the separate project The Empty House (Romanian), based on the 2021 census.
  • What POP107D counts. Population by registered legal domicile, not de facto residence. Someone can be domiciled in one locality while living, working, or residing in another, including abroad.
  • The quadrants. "Paradox" = housing stock grew, population fell. "Expansion" = both grew. "Contraction" = both fell. "Pressure" = population grew faster than housing stock (the rarest category, 10 localities, typically larger cities where the old stock cannot keep pace).
  • Cross-check. LOC101B's total 2025 housing stock (9,944,989) is within 0.2% of the total dwelling count from the 2021 Population and Housing Census (9.93 million) — different years and methodologies, but a reasonable independent confirmation of the order of magnitude.
  • No causal claim. The correlation between rising housing and falling population does not say why it happens in any given locality; the "Why this happens" section above lists plausible mechanisms, not conclusions.
  • The map. The boundaries of the 3,181 administrative-territorial units are reused from the "When Did Modernity Reach You?" project (source: geo-spatial.org and ANCPI, CC BY-SA 4.0 licence), matched to the statistics office's data through the SIRUTA code.

The data can be consulted directly in TEMPO-Online, matrices LOC101B and POP107D.

↑ Top