ImmigrationDBVerified immigration data

German rent tracks density and jobs — not crime, not clean air

By · Published · Updated

Question: what does a higher rent in Germany actually go together with?

Answer: density and size, and a labour market — not crime, and not dirty air. Across Germany's 399 housing-market districts, advertised rent ranks with population density at +0.47 and with district population at +0.37. It moves against unemployment at −0.28. And the two things people most often assume it prices in barely register: recorded crime at −0.11 and annual PM2.5 at −0.12.

None of these is an explanation. They are bivariate rank correlations between published values, and the characteristics are entangled with each other. What they can do is dismantle a few assumptions.

What we measured

The rent measure needs naming exactly, because it is not the number on a German tenant's existing lease. It is the BBSR Wiedervermietungsmieten inserierter Wohnungen — the average advertised rent for re-letting an existing flat, drawn from property portals and newspaper listings, excluding new-build. It excludes utilities and operating charges, so it is not a total housing bill either. It is what somebody searching today is quoted.

That series is published to district level, and so is the German recorded-crime indicator used here — both are BBSR compilations. The 2,058 German city rows in the packet therefore carry only 399 distinct rent values: a median of four cities share one figure, and the largest district covers 27. Correlating city by city counts the same pair of numbers up to 27 times, inflating the apparent sample without adding a single independent observation. So every coefficient below is computed once per district, against the median of its member cities for the city-level characteristics.

That choice is not cosmetic. It changes two of the six results materially:

Germany is the only country in the packet with enough co-published rent, population, density, unemployment, crime, air-quality and foreign-share data to support this at scale. The other 42 rent rows in the file come from seven different national systems and are not used here.

The median district advertises €9/m² per month; the range runs from €6 to €21. Rents are dated 2024. The other characteristics are latest published values and are not synchronised to that year — a real limitation, stated again in the limits below.

What the data shows

Density is the clearest signal: +0.47 across 395 districts. District population follows at +0.37. Denser, larger places rank higher on advertised rent, which is neither surprising nor useless — it says the market tier a place belongs to is broadly legible from its size.

The figure carries the qualification the coefficient cannot. A positive rank correlation of 0.47 is a tendency, not a formula. Districts of comparable density sit several euros apart, and knowing a district's density will not tell you its rent.

The most interesting result is what does not appear. Recorded crime correlates with rent at −0.11 — close to nothing. Measured at city level the same underlying numbers give −0.26, more than twice as strong, purely because each district's single crime figure and single rent figure are repeated across its member cities. Both variables being district-level compilations makes this the worst case for that error, and correcting it removes most of the apparent relationship.

Annual PM2.5 behaves similarly: −0.12 across 392 districts, where a lower value means cleaner air. Expensive districts are not, on average, notably dirtier or notably cleaner.

Unemployment moves the other way from the submitted reading. At district grain it is −0.28 across 382 districts, nearly twice the city-level figure and the strongest negative association in the set. More expensive districts tend to have lower unemployment. That is the one "what rent prices in" candidate here with a real signal behind it, and it still does not establish direction: strong labour markets and expensive housing are two faces of the same regional economies.

Foreign-population share shows +0.47, but across only 120 districts — under a third of the sample. It is reported with its n rather than treated as a general result.

Where it breaks down

The national median is a better tool for a named place than any coefficient. Against the €9 district median:

The Munich district advertises €21/m², 133% above the median, at 7.9 µg/m³ of PM2.5. Berlin advertises €17/m², 89% above, at 11.3 µg/m³. Frankfurt am Main and Hamburg sit between them. Berlin's air reading is markedly worse than Munich's while its rent is markedly lower — whatever separates these two markets, air quality is not it.

At the other end, districts advertising €6/m² — a third below the median — are concentrated in eastern Germany and in thinly populated western districts. Places like Gera and Görlitz sit in this band. Their low rents come with the profile the coefficients predict: low density, small population, and — reliably — higher unemployment than the expensive districts.

Two cautions about these tables. The published values are whole euros, so many districts tie at €6, €7 or €8, and the ordering within a tie carries no information. And every city inside a district shares its figure, so a table row describes a market area, not a town.

What this means if you are moving

Use density and size to place a market, not to price it. They tell you which tier a district belongs to; they will not tell you what a flat costs.

Do not read rent as a safety or air-quality signal in either direction. At −0.11 and −0.12 those relationships are effectively absent at the grain the data is published at, and the stronger-looking city-level versions are an artefact of repeated values. If safety or air quality matters to you, read those series directly for the place you are considering.

Take the unemployment relationship seriously as context rather than as a rule. Cheaper districts do tend to have weaker labour markets, and a low advertised rent in a district with high unemployment is describing a coherent regional situation, not a bargain that others have missed.

Finally, remember what the number is. Advertised re-letting rent excludes utilities, heating and operating charges, and it describes what is on the market rather than what sitting tenants pay. Check the city's own Mietspiegel where one exists — in the largest cities the two differ by a factor approaching two.

Limits

Every coefficient is bivariate. Density, size, unemployment, crime and air quality are correlated with one another, and nothing here isolates an independent effect. There is no regression, no adjustment, no effect size in euros.

The grain is a housing-market district. Neighbourhood variation is invisible, and so is variation between the towns inside one district. A district-level correlation need not hold at street level.

Coverage is uneven: foreign share pairs with only 120 districts, unemployment with 382, PM2.5 with 392, against 399 for population and crime. Missing values are not zeroes, and the characteristics are not observed in the same year as the 2024 rents.

The rent measure is advertised, charges-excluded and rounded to whole euros. It is not total housing cost, not the rent of an existing tenancy, and not a guarantee that any dwelling is available at that price.

Recorded crime depends on reporting and classification practice and is not a measure of personal risk. Annual PM2.5 is a background mean, not exposure at an address.

Method

The analysis uses the 2,058 German cities that carry the BBSR advertised-rent series and collapses them to their 399 distinct districts before any coefficient is computed. Each district contributes its published rent once; for the city-level characteristics — density, population, crime, PM2.5, unemployment, foreign share — it contributes the median of its member cities' values. Districts are weighted equally.

Spearman's rank correlation is used throughout, because the distributions are skewed, the rent values are tied at whole euros, and the question is about ordering rather than about euros per unit. Each coefficient is reported with the number of districts behind it, which varies by characteristic according to coverage.

The city-level figures shown for comparison in the "What we measured" section use exactly the same data without the district collapse, and are reported only to show the size of the distortion. They are not used for any conclusion.

The median district rent is €9/m² per month. The two tables rank districts by percentage distance from that median, calculated as (district rent / median rent − 1) × 100. Distance from the median is not a claim about over- or under-pricing, and no multivariable residual was estimated.

Sources

Frequently asked questions

What is most closely associated with rent in Germany?

Population density, at Spearman +0.47 across 395 housing-market districts, followed by district population at +0.37 across 399. Both are bivariate rank correlations: they show which districts sit high on both measures, not how many euros density adds.

Do more expensive German places have more recorded crime?

Barely a relationship either way. At the grain both figures are actually published at — the district — rent and recorded crime correlate at −0.11 across 399 districts. Measured city by city the same data gives −0.26, but that counts each district's single rent and crime value up to 27 times.

Does rent track the job market?

More than it tracks crime. Rent and the unemployment rate correlate at −0.28 across 382 districts: more expensive districts tend to have lower unemployment. It is the strongest negative relationship in the set and it does not establish that either causes the other.

Why does this article show €21 for Munich when the Munich page shows €15.38?

They are different measures. This article uses BBSR advertised re-letting rents, which give the Munich district €21/m² per month for 2024. The city page shows the official Mietspiegel reference rent of €15.38/m², which covers existing tenancies rather than new listings.

Why is the analysis reported for districts rather than cities?

Because that is the grain the data is published at. Both the rent series and the German crime indicator come from BBSR compilations produced to district level, so 2,058 city rows carry only 399 distinct values — a median of four cities share each one.

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.