Accessibility inequality across Europe: a comparison of 15-minute pedestrian accessibility in cities with 100,000 or more inhabitants

The study addresses how unequal current pedestrian accessibility conditions are across European cities. Motivated by efforts to reduce car dependence and promote sustainable modes, it focuses on active accessibility—people’s ability to reach relevant opportunities by walking. The 15-minute city concept underscores proximity to diverse local destinations, but it faces challenges: defining which opportunities are relevant for different people, variable acceptable travel times by purpose, and mismatches between perceived and physical walking times. Existing active accessibility research has largely been city-specific with limited cross-country pedestrian-focused comparisons. Recognizing accessibility’s role in distributive justice, this study evaluates within-city and between-city inequalities in 15-minute pedestrian accessibility across European cities with 100,000+ inhabitants, using two indicators: Total Destinations (cumulative opportunities) and Variety (count of different opportunity types).

Prior work has documented high car dependence and the need to prioritize active modes in sustainable urban mobility. Walkability indices (e.g., Walk Score) and recent tools (OSM-WALK-EU) facilitate cross-context assessment but have limitations in coverage (especially Europe) or scalability. Other global comparisons typically assess access to jobs or generic services using multimodal or driving networks, not specifically pedestrian access to everyday amenities. Literature in transport equity highlights unequal distribution of mobility benefits and burdens, supporting accessibility-based, justice-oriented evaluation. The 15-minute city concept emphasizes access to diverse opportunity types rather than sheer quantity. However, setting universal thresholds is problematic due to differences in perceived walking time and varying importance of destinations by trip purpose. This study contributes by offering a harmonized, replicable approach for cross-city comparison of pedestrian accessibility diversity and inequality across Europe.

Design: Place-based pedestrian accessibility analysis for cities with ≥100,000 inhabitants, evaluating both levels and inequalities within and between cities. Sample: 585 European cities across 31 countries (from an initial 864 cities; threshold based on most recent 2011–2020 Eurostat population data). Data sources: City boundaries from Eurostat/GISCO Urban Audit 2020 (city and greater city units); street and pedestrian network from OpenStreetMap (OSM) via Pandana (v0.6.1), excluding non-walkable segments; destinations from OSM via pyrosm (v0.6.0), downloaded 15 June 2021; population weights from the Global Human Settlement Layer (GHS-POP R2023) for inequality analysis; pseudo-Gini coefficients computed with PySAL (v2.1.0). H3 hexagonal grid (level 10; mean edge ~65.9 m; mean area ~15,047.5 m²) used to harmonize spatial units across cities. Destinations selection and classification: OSM POIs with keys amenity, craft, leisure, office, shop, tourism; destinations with access=no/private/customers excluded. Non-residential buildings (polygons) converted to centroids to capture major destinations (e.g., hospitals, universities) not tagged as POIs; duplicates avoided by excluding buildings inside polygons with other destination tags. To mitigate edge effects, a 1-km buffer outside each city boundary was included when selecting destinations. Destinations grouped into 10 types for Variety: (1) Education; (2) Supermarkets/markets/food shops; (3) Healthcare; (4) Sports and recreation; (5) Culture and leisure; (6) Parks/green areas; (7) Eating and drinking establishments; (8) Retail; (9) Religious; (10) Public service. Accessibility computation: Origins are all nodes of the pedestrian network (Pandana). Cumulative opportunities measure used: Ai = Σj φ(Cij), with φ(Cij)=1 if Cij≤δ, else 0, where δ is the travel-time threshold. To represent a 15-minute walk, three constant walking speeds were used: 0.7, 0.9, 1.1 m/s, corresponding to radii of 630, 810, and 990 m. Accessibility was computed for each speed and averaged to yield the final score per origin node. Two indicators were produced per origin: (a) Total Destinations (count of all accessible destinations), and (b) Variety (number of distinct destination types accessible; range 0–10). Node-level results were aggregated to H3 cells by assigning the value of the node nearest the hexagon centroid. The final dataset comprises 4,347,078 hexagon observations across all cities. Inequality metrics: Two pseudo-Gini coefficients were calculated for each city and indicator: Territory-based Gini (T-Gini) across hexagons (spatial inequality) and Population-based Gini (P-Gini), weighting hexagons by estimated residents from GHS-POP (inequality across people). Entropy was also tested but highly correlated with Variety (r²≈0.955); thus, Variety was retained as the primary diversity measure.

• Cross-country variation: Mean Total Destinations ranged from 17.5 (Sweden) to 136.4 (Switzerland). Countries with high means included Ireland (90.9), Austria (80.0), Luxembourg (75.3); low means included Finland (20.5), Slovenia (24.2), Hungary (27.5). Variety means were lowest in Sweden (2.9) and highest in Ireland (8.5); high Variety also in Malta (7.9), Switzerland (7.9), Luxembourg (7.2); low Variety in Finland (3.6), Bulgaria (3.8), Hungary (4.0).
• Across cities: City-level mean Total Destinations ranged from 3.3 to 593.4 with a positively skewed distribution; approximately half of cities had means below 50 and fewer than 10% above 100. Variety ranged from 1.18 (Kuopio, FI) to 9.90 (Islington, UK) with an approximately normal distribution (mean ≈5.91). About 6.8% of cities had very low mean Variety (≤3), and ~11% had very high mean Variety (≥8).
• Within-city patterns: Total Destinations typically showed a single central hotspot, sometimes polycentric, with very large intercity differences in hotspot magnitude (e.g., ≈70 in Borås, SE; ≈220 in Lahti, FI; >2,350 in Greater Amsterdam, NL). Variety revealed additional polycentric areas beyond central hotspots, indicating locations with diverse accessible opportunities even where total counts are modest. All cities had at least one location with maximum Variety (10 types), but citywide mean Variety remained far from the maximum in most cases.
• Relationship between indicators: Variety increased with Total Destinations following an upward exponential decay (r²≈0.662), indicating diminishing marginal returns of additional destinations for diversity; some cities had high totals but lower-than-expected Variety (concentration in fewer types).
• Density relationships: Total Destinations rose roughly linearly with population density, while Variety followed an upward exponential decay with diminishing returns (r²≈0.605), suggesting that beyond a threshold (~7,500 inhabitants/km² or 75 inhabitants/ha), additional density yields smaller Variety gains; urban structure and local economy likely also shape Variety.
• Inequality: For both indicators, P-Gini < T-Gini, indicating that population distribution attenuates spatial inequality. Mean Gini dropped from 0.665 (T-Gini) to 0.541 (P-Gini) for Total Destinations, and from 0.333 (T-Gini) to 0.153 (P-Gini) for Variety. Variety inequality (both T-Gini and P-Gini) was systematically lower than Total Destinations inequality across all cities.
• Correlations with inequality: Variety had a strong negative association with inequality (T-Gini r²≈0.971; P-Gini r²≈0.656), while Total Destinations had weak associations (T-Gini r²≈0.207; P-Gini r²≈0.083). Larger, denser cities generally exhibited higher Variety and lower inequality, with more variance among smaller/less dense cities; outliers were identified primarily among cities below 500k population.
• Overall: European cities are not yet 15-minute cities; significant within-city inequalities persist, but cities with higher Variety tend to be more equal. Increasing the diversity of accessible opportunity types can improve accessibility and reduce inequality without requiring extremely high destination counts or densities.

The study’s central question—how unequal are pedestrian accessibility conditions across European cities—is addressed through two complementary indicators. Total Destinations effectively reveals within-city hotspots but varies so widely across cities that it is less useful for cross-city comparisons. Variety, by focusing on accessible types of opportunities, enables meaningful comparisons across cities and within cities, revealing polycentric clusters of diverse access that can exist even where total counts are modest. The diminishing returns between Total Destinations and Variety imply that after a baseline of accessible opportunities is reached, further increases in total counts do not proportionally increase diversity. The density analysis shows that while some density is necessary for good pedestrian accessibility, Variety exhibits saturation effects beyond roughly 75 inhabitants/ha, elevating the importance of urban structure, land-use mix, and the local economy. Inequality findings—systematically lower for Variety than for Total Destinations and lower for P-Gini than T-Gini—suggest that both the distribution of people and a more polycentric, mixed-use structure mitigate disparities. These results are salient for equity-focused planning: prioritizing the distribution of diverse everyday destinations can yield more equitable pedestrian access. The findings support the 15-minute city’s emphasis on diverse local opportunities while highlighting limitations of a single threshold and the need to consider purpose-specific tolerances for walking.

This paper contributes a harmonized, scalable method to compare 15-minute pedestrian accessibility and inequality across 585 European cities using open data and open-source tools. It shows that: (1) European cities are far from universal 15-minute conditions; (2) within-city inequalities are substantial but lower where Variety is higher; (3) Variety provides a robust, comparable metric across cities; (4) density supports accessibility but exhibits diminishing returns for Variety beyond ~75 inhabitants/ha; and (5) population-based inequalities are consistently lower than territory-based ones. Policy implications include emphasizing the spatial distribution of a diverse set of everyday opportunities and creating multiple mixed-use centralities to enhance equity. Future research should: adopt varying time thresholds by opportunity type; integrate cycling and multimodal accessibility; incorporate micro-scale pedestrian infrastructure and barriers; refine city boundary definitions; and extend the approach beyond Europe.

• Threshold and scale effects: Results depend on a fixed 15-minute threshold and the chosen spatial aggregation (H3), reflecting the Modifiable Areal Unit Problem (MAUP).
• Walking speed assumptions: Conversion from time to distance used constant speeds (0.7, 0.9, 1.1 m/s), potentially overstating accessibility for slower walkers (e.g., many older adults) and ignoring slope or weather.
• Data completeness: Reliance on OSM POIs/buildings may omit or misclassify opportunities; coverage varies across cities.
• City boundary heterogeneity: Eurostat/GISCO ‘city’ vs ‘greater city’ definitions are not uniform across countries, potentially biasing comparisons.
• Network detail: Micro-scale pedestrian features (sidewalks, crossings, gradients, safety) were not represented; results should be interpreted as potential rather than realized pedestrian accessibility.