Equitable by position

18 tram lines and 116 day bus lines, with 22 more at night. On a weekday they make about 10,400 trips.

The network is radial. The tram lines meet in the centre and run out along the main roads to the large housing estates. The estates of Piątkowo and Winogrady in the north are linked to the centre by a rapid tram on its own track, built in the 1990s, which carries lines 12, 14, 15 and 16. The outer districts are served by buses, many of them feeding the trams.

Most vehicles carry door counters, which record how many people board and alight at each stop: 45% of the 67.9 million stop visits from January to October 2025.

Lines and trips: ZTM Poznań timetable (GTFS), October 2025, Tuesday to Thursday, day lines; night lines are numbered 2xx. Counts: vehicle location and passenger counting records of the PEKA system. On the rapid tram: Gadziński & Radzimski (2016).

Deprivation in Poznań is highest in the tenement districts of the pre-war core and inner ring, such as Łazarz and Stare Miasto, and in one large estate, Nowe Winogrady Wschód. It is lowest in the large estates of the north.

We measure it by registered unemployment and social assistance per 1,000 residents of each of the 42 neighbourhoods, in the city's 2019 and 2021 rounds; the high scores of Główna and Krzesiny come from single institutions. Counted by residents, nearly three quarters of the residents of the most deprived third of stop catchments also live in the third with the best access to jobs.

From the median stop of the most deprived third, a rider reaches 32% of the city's jobs within 30 minutes in the morning peak. From the least deprived third, 9%.

The advantage holds in the early morning, in the evening and on Sundays. By the usual standard of transport equity, access to jobs is progressive.

Median stop of each third of the 428 compared stop catchments, ranked by deprivation. Access by the October 2025 timetable, morning peak 07:00–09:00, from a modelled 200 m grid of jobs.

Deprived stops have no more service than other stops at the same distance from the centre and in the same direction. Their advantage is where they are.

To see how much of it is geography, keep the network and move deprivation. Re-allocate the same 42 neighbourhood scores so that deprivation rises with distance from the centre, as in cities where poverty lies at the edge.

The same network would then give the most deprived third 3.8% of the city's jobs within 30 minutes and the least deprived 42.5%. Where deprivation does lie at the edge, in five neighbourhoods south of the centre served by buses alone, the network gives it no advantage: beyond 6 km they reach 1.9% of the city's jobs within 30 minutes, against 4.9% for the rest of the periphery.

Median stop of the most and the least deprived third. Mirror allocation: the most deprived score to the most distant neighbourhood, and so on.

Riders from more and less deprived areas use the lines on offer much alike, so the network decides where they go: in two streams that cross at the centre.

Riders from the least deprived third ride north–south, along the rapid tram from Piątkowo and Winogrady to Rataje. Riders from the most deprived third ride west–east, on shorter routes from Ogrody, Jeżyce and Łazarz to Śródka. Where they get off, in the centre, the streams meet.

It looks like segregation, but most of it would appear wherever deprivation lay: scattered at random 999 times, or turned 47 ways around the centre, the same scores give about as much separation between corridors. Only the riders of the least deprived estates keep apart on their own lines more than that geography explains.

Once position on the network is held fixed, we detect little difference in how much, when or how responsively transit is used.

Each row is a gap between more and less deprived stop catchments. Switch from the raw gap to the gap net of position on the network, then net of land use and population too.

Most gaps vanish once position is held fixed. The one that survives with confidence, the deprivation of destinations, depends on where the lines run, so it belongs to places. That is not the same as no difference: when we injected a threefold difference in early boardings into the counts, these comparisons usually missed it. Door counts by area show the pulse of places; the pulse of people needs data on riders.

Nine of the 22 measures in the paper. Slopes in standard deviations per standard deviation of catchment deprivation, 95% intervals. A filled dot: the gap passes a false-discovery-rate correction across the 22 measures.

The large housing estates built under state socialism housed a broad social mix, and most have aged without declining. In Poznań, as in other Polish cities, deprivation concentrated instead in parts of the pre-war inner city.

Large housing estates

Built under state socialism at the edge of the city, they still house a broad mix of residents. On average they have aged without declining, though their paths diverge locally. Kovács & Herfert (2012); Temelová et al. (2011).

The pre-war inner city

Tenement districts such as Łazarz, Wilda and Jeżyce combine dense, mixed uses and good transit with old housing, much of it in poor repair. Social-assistance records have placed Poznań's strongest concentrations of poverty there. Kotus (2006); Weltrowska & Kisiała (2014); Kubeš & Kovács (2020).

Slow regeneration

Since 2006 the city has regenerated parts of the core, and prices and rents have risen. Gentrification has been slow and filtered by institutions: municipal housing and protected tenancies keep many deprived residents in the core. Ciesiółka & Maćkiewicz (2020); Bernt & Ogrodowczyk (2025).

Elsewhere

In many Western cities the gentrification of transit-rich cores has gone together with the suburbanisation of poverty. Central deprivation is not only post-socialist, though: in Toronto, disadvantaged areas have had better transit access than the rest of the region. Hochstenbach & Musterd (2018); Foth et al. (2013).

By the usual standard of transport equity, Poznań scores well. But its advantage is positional, and it is not enough on its own.

Deprived areas have far better access than the least deprived, and remain deprived. And the advantage lasts only while deprived residents stay where the network serves them well. Move some of them out:

A stylised illustration, not a forecast. Means over the residents of the most deprived third after the move. Access-poverty line: half the median access, 13.9% of the city's jobs within 30 minutes.

Keep the advantage

Keeping the municipal housing stock in the core may help protect the access of the residents who need it most. This study does not test it.

A floor for the edge

Standards that guarantee everyone a minimum level of access point to the deprived enclaves the network serves least. Martens (2016); Pereira, Schwanen & Banister (2017).

Plan works with riders in mind

Riders from more and less deprived areas load different parts of the network. Closing the northern rapid tram falls more on riders from the least deprived areas; closing the inner city's west–east axis or the lines to the main station falls more on riders from deprived ones.

Compact, mixed, well-served cores are what sustainable-city policy asks for, from transit-oriented development to the 15-minute city. Poznań shows that such a core can also be where deprived residents are best served by public transport, as long as they can stay in it.

42 neighbourhoods, 607 stop complexes, ten months of door counts and one timetable.

Every address goes to the stop complex it reaches soonest on foot, up to 400 m, and each stop takes the mean deprivation of the residents around it. Where riders get off is estimated on each counted trip; spread over a stop's lines by their departures, the same estimate gives what the network offers. Door counts see areas, not people, and only those who ride.

The analysis code is on GitHub. The door counts and the address register of residents cannot be shared; the timetable, the topographic database, the revitalisation indicators and OpenStreetMap are public.

• Marcin Wozniak

  Faculty of Human Geography and Planning, Adam Mickiewicz University, Poznań.

  wozniak2.github.io

• Gabriele Filomena

  Geographic Data Science Lab, University of Liverpool.

  PedSimCity