The objective of this chapter is to develop an experimental typology of intermediary cities. First, this chapter proposes the indicators selected for the functions of intermediary cities identified using the definition introduced in Chapter 2 and illustrates how these key functions are performed in different parts of the EU and OECD countries. Second, the chapter analyses the distribution of functions across cities and co-occurrence of functions (e.g. mono-functional vs. pluri-functional cities, clusters of functions) for selected EU countries with available data, leading to the identification of five types of intermediary cities. Lastly, this chapter analyses the diverse characteristics in terms of territorial, governance institutional and economic contexts found in the five identified types of intermediary cities.
Unlocking the Potential of Intermediary Cities for Regional Development
Abstract
In this section, a set of indicators are identified to map how the defined eight functions are operating in different parts of the EU and OECD countries. These indicators are carefully selected by balancing criteria of measurability and practical applicability (Table 5.1).
Source: Author’s own elaboration.
5.1.1. Business
The indicator associated with the Business function is the presence of Knowledge Intensive Business Services (KIBS), expressed as the share of employment in KIBS, as measured in 2021. The NACE Rev 2 division codes corresponding to KIBS activities are: Computer programming, consultancy and related activities (62), Information service activities (63), Legal and accounting activities (69), Activities of head offices; management consultancy activities (70), Architectural and engineering activities; technical testing and analysis (71), Advertising and market research (73), and Other professional, scientific and technical activities (74). The employment share in KIBS is reported in relation to the national average, which corresponds to 100% (Figure 5.1). Data were collected from National Statistical Offices at municipal level and subsequently aggregated for FUAs, for the following countries: Austria, Finland, Germany, Italy, Latvia, Portugal, Spain and Sweden.
The distribution of intermediary cities with the highest and lowest employment share in KIBS within each country is rather scattered. Overall, approximately 15% of the analysed intermediary cities (27 over 183) are performing better than the national average on this indicator. Of the remaining intermediary cities, a few are performing only marginally worse than the national average, and the rest (more than 70%) are characterised by an employment share in KIBS that is only a small fraction of the national average. The intermediary cities with the lowest values of employment share in KIBS, less than half of the national average, are distributed across Germany, Latvia and Spain. The intermediary cities where the presence of KIBS appears to be the highest are in Spain (Santiago de Compostela) and in Italy, with examples from both Northern (Pavia, Treviso), and Central and Southern regions (Cosenza, Pisa, Potenza). The large variation in the Business function in some countries highlights the heterogeneity of intermediary cities, which stems from multiple factors, including their proximity to larger cities and differences in institutional settings (e.g. federal or unitary states).
Employment share in Knowledge Intensive Business Services relative to the national average (%), 2021
Note: Employment in Knowledge Intensive Business Services is defined as employment by place of work in the subset of knowledge-intensive service activities corresponding to business services, adapted from the Eurostat definition of knowledge-intensive services (Eurostat, n.d.[1]). The NACE Rev 2 divisions included are: 62, 63, 69, 70, 71, 73 and 74. Division 72 (Scientific research and development) is excluded to avoid overlap with the knowledge function. Values are expressed relative to the national average which is indexed to 100. EU-27 countries covered: Austria, Finland, Germany, Italy, Latvia, Portugal, Spain and Sweden.
Source: Calculations based on employment-by-industry data from national sources and OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.2. Knowledge
The indicator associated to the Knowledge function is the number of students per capita enrolled in higher education institutions (HEI), relative to the national average. The education programmes considered are those from level 5 to level 7 according to the International Standard Classification of Education (ISCED)1 classification: Short-cycle tertiary education (ISCED 5), Bachelor’s or equivalent level (ISCED 6), and Master’s or equivalent level (ISCED 7). The number of students enrolled in HEI is reported over the total resident population (per capita), and in relation to the national average, which corresponds to 100% (Figure 5.2). Data from the European Tertiary Education Register are available for 17 of the 23 EU-27 countries with intermediary cities: Austria, Bulgaria, Croatia, Czechia, Estonia, Finland, Germany, Hungary, Ireland, Italy, Latvia, the Netherlands, Poland, Portugal, the Slovak Republic, Spain and Sweden. Data are available also for non-EU OECD countries such as Norway and the United Kingdom.
Out of 327 analysed intermediary cities, 286 of them are located in EU-27 countries, 40% of which do not host any HEI. The countries where all intermediary cities host at least one HEI are Croatia, Finland, Latvia and the Slovak Republic. Austria, Ireland and Norway have only one intermediary city and they also host at least one HEI. Overall, the pattern of intermediary cities potentially providing the Knowledge function is rather diverse across countries. For instance, while Germany and the United Kingdom have a similar number of intermediary cities (44 and 40 respectively), in Germany 75% of them host at least one HEI whereas the share in the United Kingdom goes down to 25%. Similarly, while Italy and Spain have 59 and 55 intermediary cities respectively, 62% of the Italian intermediary cities host at least one HEI, against only 31% in Spain. At the same time, 15 out of the 17 Spanish intermediary cities hosting at least one HEI have a number of HEI students per capita higher than the national average, while in Italy this proportion goes down to 50%.
Note: Number of students enrolled in higher education institutions in programmes classified as ISCED 5 to 7 in 2021 (or latest year available). Values are expressed per capita and relative to the national average which is indexed to 100. EU-27 countries covered: Austria, Bulgaria, Croatia, Czechia, Estonia, Finland, Germany, Hungary, Ireland, Italy, Latvia, the Netherlands, Poland, Portugal, the Slovak Republic, Spain and Sweden. Non-EU OECD countries covered: Norway and the United Kingdom.
Source: Calculations based on OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.3. Transport
The indicator associated with the transport function is based on the road transport performance methodology developed by DG REGIO (2019[2]). The indicator is defined as the accessible population (people reachable within a 90-minute drive) divided by the nearby population (people living within a radius of 120 km). If the entire nearby population can be reached quickly, the transport network is performing well. The indicator is available for all EU-27 countries with intermediary cities except for Cyprus.
Transport performance also depends on the level of urbanisation of the countries and their road density: for instance, Belgium and the Netherlands, which are relatively small (in land surface) and highly urbanised countries, score the highest across Europe. Intermediary cities reflect these territorial patterns: the best performing intermediary cities in the Transport function can be found in Belgium, France, Germany, the Netherlands, Spain and the United Kingdom (Figure 5.3)
Note: Road transport performance is defined as the ratio between accessible population (people reachable within a 90min drive) and nearby population (how many people live within a 120km radius). Road network is measured in 2016, population is measured at grid level in 2011. EU-27 countries covered: Austria, Belgium, Bulgaria, Croatia, Czechia, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, the Slovak Republic, Spain and Sweden. Non-EU OECD countries covered: Norway, Switzerland and the United Kingdom.
Source: Calculations based on DG REGIO (2019[2]) and OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.4. Housing
To capture the role that intermediary cities can play in providing housing, the indicator associated with this function is the share of working population residing in a city and that work outside the city.
The indicator is available for 13 of the 23 EU-27 countries with intermediary cities: Austria, Belgium, Czechia, France, Germany, Hungary, Italy, Latvia, the Netherlands, Portugal, the Slovak Republic, Spain and Sweden. Among the analysed EU countries, Belgium, Germany, Italy, the Netherlands, Portugal and Spain have the intermediary cities with the highest share of working population commuting outside the city boundaries (above 50%). However, countries show different patterns. For instance, in almost 74% of Dutch intermediary cities, the share of working population commuting outside the city boundaries is higher than 30%, while only 36% of Spanish intermediary cities have a share of working population commuting outside the city boundaries that is above 30% and these intermediary cities are concentrated along the Mediterranean coast (Figure 5.4). This may imply that, in general, the Dutch urban system provides more flexibility in terms of selecting housing locations, partly thanks to the availability of housing as well as the proximity and connectivity across cities.
Note: The share of working population with a job outside the city is obtained from municipal level commuting flow data in 2023 (or latest year available), counting the number of people commuting to a municipality outside of their city, and rescaling by the total size of the working population in that city. EU-27 countries covered: Austria, Belgium, Czechia, France, Germany, Hungary, Italy, Latvia, the Netherlands, Portugal, the Slovak Republic, Spain and Sweden. Non-EU OECD countries covered: Norway and Switzerland.
Source: Calculations based on OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.5. Culture
The indicator associated with the Culture function is the number of cultural amenities per capita in an intermediary city. It is based on data from Overture Maps (2024[3]), which provides broad geographic coverage across intermediary cities in the EU. The indicator includes the following types of cultural points of interest: landmarks and historical buildings, libraries, museums, cinemas, theatres and performance venues, musical venues, operas and ballets. While the indicator captures the presence of a wide range of cultural amenities, it does not encompass all dimensions of a city’s cultural offer. Certain categories, such as cultural centres, are not included, nor does the indicator capture broader cultural infrastructure, such as studios or workshop spaces. In addition, it does not reflect differences in the quality, scale or intensity of use of cultural amenities: as a result, it may provide only a partial picture of a city's cultural ecosystem. To mitigate the effects of cross-country differences in data coverage and reporting intensity, the indicator is scaled relative to the national average. This approach highlights the position of intermediary cities within their national context.
Across all analysed countries, 27% of the intermediary cities assessed are providing a number of cultural amenities per capita that is at least equal to the national average. However, the indicator reveals a heterogeneous picture across the EU. The analysed countries without any intermediary city scoring above the national average are Estonia, Finland, Latvia, Lithuania and Sweden (Figure 5.5). In Portugal, the only intermediary cities scoring better than the national average are located in the country’s archipelagos: Ponta Delgada (Azores) and Funchal (Madeira).
Note: Number of cultural amenities is reported per capita and compared to the national average, which is set at an index of 100. Cultural amenities are measured in 2023 (or latest year available) and are defined as the sum of landmarks and historical buildings, libraries, museums, cinemas, theatres and performance venues, musical venues, operas and ballets, as reported by Overture Maps. EU-27 countries covered: Austria, Belgium, Bulgaria, Croatia, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, the Slovak Republic, Spain and Sweden. Non-EU OECD countries covered: Norway, Switzerland and the United Kingdom.
Source: Calculations based on Overture Maps Foundation (2024[3]) and OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.6. Retail
The indicator associated with the Retail function is the number of shopping centres per capita in each intermediary city, compared to the national average: a value higher than 100% indicates that the intermediary city offers more shopping centres per capita than the national average. The indicator is computed using point-of-interest data from Overture Maps (2024[3]), which was selected for its wide geographic coverage and ease of comparability. Shopping centres are often located outside of city centres, complementing traditional high-street retail. As such, the indicator provides a consistent measure of access to one dimension of retail services across intermediary cities, but it does not reflect differences in their quality, scale or levels of use. The indicator is available for all EU-27 countries with intermediary cities except for Cyprus.
Overall, more than 60% of the analysed intermediary cities perform better than the national average. While each of the analysed countries has at least one intermediary city with a score higher than 100%, the occurrence of intermediary cities where the number of shopping centres per capita is considerably higher than the national average is more frequent in Central and Eastern Europe, with approximately 91% of intermediary cities in Czechia, more than 96% of intermediary cities in Romania, and all five intermediary cities in Croatia (Figure 5.6).
Note: : Shopping centres are shown per capita and compared to the national average, which is set at an index of 100. Shopping centres are measured in 2023 (or latest year available). EU-27 countries covered: Austria, Belgium, Bulgaria, Croatia, Czechia, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, the Slovak Republic, Spain and Sweden. Non-EU OECD countries covered: Norway, Switzerland and the United Kingdom.
Source: Calculations based on Overture Maps Foundation (2024[3]) and OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.7. Health
The indicator associated with the Health function is the number of hospitals per capita in each intermediary city, relative to the national average. It is available for 18 of the 23 EU-27 countries with intermediary cities: Austria, Belgium, Bulgaria, Croatia, Czechia, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Lithuania, Latvia, the Netherlands, Portugal, Spain and Sweden. Outside the EU-27, this indicator is also available for Norway and Switzerland in the OECD. Among the intermediary cities analysed in EU-27 countries (238), almost half (47%) have a number of hospitals per capita that is equal to or above the national average (Figure 5.7). Of these intermediary cities performing at or above the national average, one third are intermediary cities with populations between 50 000 and 100 000 inhabitants. Czechia, France, Italy, the Netherlands, Spain and Sweden have at least one intermediary city that provides healthcare services almost twice and up to three times the respective national average. Almost 42% of these intermediary cities where the number of hospitals per capita is considerably above the national average, are intermediary cities with populations between 50 000 and 100 000 inhabitants, indicating the prominent role of these cities within the national health and medical systems.
Note: The number of hospitals is measured in 2023 (2024 for Germany) and is shown per capita and compared to the national average, which is set at an index of 100. EU-27 countries covered: Austria, Belgium, Bulgaria, Croatia, Czechia, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, the Netherlands, Portugal, Spain and Sweden. Non-EU OECD countries covered: Norway and Switzerland.
Source: Calculations based on Eurostat GISCO (2023[4]), Statistical Offices of the Federation and the Länder (2024[5]) and OECD Regions, Cities and Local Areas database http://oe.cd/geostats.
5.1.8. Government
The indicator for the Government function is public employment expenditure, defined as the compensation of employees paid by municipal-level governments. The indicator is computed as the population-weighted average of municipal public employment spending. It is reported per capita and compared to the national average, which is set at an index of 100. The indicator is available for 15 of the 23 EU-27 countries for which we identified intermediary cities: Austria, Croatia, Czechia, Estonia, Finland, France, Hungary, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, Spain and Sweden.
While this indicator does not capture the presence of regional offices of national or international ministries and agencies in an intermediary city, it provides useful insights into the responsibilities and functions delegated to the municipal governments included in the corresponding city, relative to the national average. Out of the countries included in the analysis, Finland, Latvia, Lithuania, Poland and Sweden have more than half of their intermediary cities with a level of municipal public employment expenditure per capita equal or higher than the national average. For instance, 76% of Polish intermediary cities are above the national average (Figure 5.8).
Note: Public employment expenditure data refers to the compensation of employees from the 2008 System of National Account (SNA 2008) by municipal level governments only. It is defined as the population-weighted average (FUA level) of municipal public employment spending and is expressed per capita and compared to the national average, which is set at an index of 100. This data is from 2022 or the latest year available. EU-27 countries covered: Austria, Croatia, Czechia, Estonia, Finland, France, Hungary, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, Spain and Sweden. Non-EU OECD countries covered: Norway.
Source: Calculations based on OECD MUNIFI Database.
5.2.1. How are functions distributed across intermediary cities and larger cities?
In order to understand how different functions are distributed across intermediary cities and larger cities, Table 5.2 provides an overview of intermediary cities that have available data for the seven functions Business, Knowledge, Transport, Housing, Culture, Retail and Health. The Government function is not included in this overview because of two main reasons. First, while the indicator of the Government function is useful to capture within-country differences, it may be more difficult to interpret when comparing intermediary cities across countries, due to differences in the governance structure (e.g. centralised vs decentralised). Second, the country coverage of the indicator of the Government function is somewhat limited and including it in the analysis would have further reduced the sample size.
After retaining only cities that have non-missing values across all functions, the final sample consists of 173 intermediary cities across Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden. Intermediary cities in Portugal and Spain that are not in mainland Europe are excluded from the sample, as they feature vastly different profiles. The final sample, despite covering only part of the intermediary cities identified in EU-27, provides sufficient variation in terms of country coverage across different regions of Europe, and it can be used for a pilot exercise to provide preliminary insights on how different functions are distributed and co-occur in intermediary cities.
Table 5.2 details the count and percentages of intermediary cities that fall into the top and bottom quintiles of each of the assessed functions.
While intermediary cities are in general less likely to offer business services compared to larger cities, some intermediary cities are doing extremely well in the Business function, especially cities in the North of Italy such as Pisa, Trento and Vicenza. Overall, intermediary cities have a more prominent Housing function that larger cities. This finding is consistent with the fact that larger cities feature larger labour markets and hence can offer more opportunities for employment. The Knowledge function tends to be rather scattered: intermediary cities are much more likely to not host any higher education institution and are thus overrepresented in the bottom quintile for the corresponding indicator. Yet, there are intermediary cities that score very high in terms of students per capita, e.g. Santiago de Compostela, Salamanca in Spain, and Caserta and Pisa in Italy.Regarding the remaining functions (Culture, Retail and Health), intermediary cities are more likely to be “top-performers”, compared to larger cities.
Number and share of cities in the top and bottom quintiles, by function and city size
Note: Table displays counts and fractions of cities falling into top or bottom quintile for intermediary cities and larger cities (cities with population above 250 000 inhabitants). Quintiles are computed separately for each function on the pooled set of intermediary cities and larger cities. Functions measured in years 2023 or latest available. Sample of cities consists of 173 intermediary cities and 115 larger cities across Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden.
Source: Analysis based on DG REGIO (2019[2]); Eurostat GISCO (2023[4]); employment-by-industry data from national sources; OECD Regions, Cities and Local Areas database http://oe.cd/geostats; Overture Maps Foundation (2024[3]); Statistical Offices of the Federation and the Länder (2024[5]).
5.2.2. How do functions co-occur in intermediary cities?
Table 5.3 is based on the same country sample used for Table 5.2 (Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden) and it focuses on how the seven functions Business, Knowledge, Transport, Housing, Culture, Retail and Health interact with each other in intermediary cities i.e. how more or less likely it is that two functions are present in the same intermediary city. The numbers reported in the table correspond to correlation coefficients between pairs of functions, ranging from -1 (the two functions tend not to be in the same intermediary city) to 1 (both functions tend to have a strong presence in the same intermediary city).
The results reported in Table 5.3 show that intermediary cities with a strong presence of business services tend to feature many enrolled students per capita, which is the indicator for the Knowledge function. Santiago de Compostela (Spain), Pisa, Caserta, Lecce, Pavia and Cosenza in Italy, and Linköping (Sweden) are top ranked on both Business and Knowledge functions.
The results indicate a positive correlation between the Business and Culture functions, supporting the argument that cultural amenities act as a broader driver of cultural and creative ecosystems, even if the relationship is somewhat less strong than between the Business and Knowledge functions. Overall, intermediary cities with many cultural amenities tend to also host more amenities of other types and points of interest.
Intermediary cities where a significant portion of the workforce commutes to jobs outside the city tend to have a well-developed road network. These intermediary cities are frequently situated along the coast, as seen by the examples of Aveiro in Portugal and the Spanish cities of Reus, Alcoy, Elda, Sagunto and Cartagena. Top-performing intermediary cities in both the Housing and Transport functions are those located near major metropolitan centres. Notable examples include Guadalajara (close to Madrid) and Igualada (close to Barcelona).
The data also show a significant negative correlation between the Business and Transport functions: intermediary cities with a strong presence of business services tend to have lower transport performance. This pattern may also be related to the fact that intermediary cities with a strong Business function tend to be located further from larger cities, where transport performance is somewhat lower.
Correlation matrix of functions in Intermediary Cities
Note: Functions measured in year 2023 or latest. Sample of cities consists of 173 intermediary cities in Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden. Stars indicate significance levels, with * p<0.05, ** p<0.01, *** p<0.001.
Source: Analysis based on DG REGIO (2019[2]); Eurostat GISCO (2023[4]); employment-by-industry data from national sources; OECD Regions, Cities and Local Areas database http://oe.cd/geostats; Overture Maps Foundation (2024[3]); Statistical Offices of the Federation and the Länder (2024[5]).
5.2.3. The five types of intermediary cities
While correlations provide insights about which functions tend to move together, they don't necessarily paint a clear picture of what “types” of intermediary cities there are. To better understand the co-occurrence of functions, intermediary cities can be clustered to obtain groups (or clusters) of cities sharing a similar profile in terms of functions provided and that can be interpreted as different “types”.
This exercise is based on the same sample of intermediary cities used for Table 5.2 and Table 5.3, which are grouped (k-means clustering) based on the functions that they provide. The method used (k-means clustering) ensures that the intermediary cities belonging to the same group are the most similar possible and it is implemented so to ensure robust results i.e. reducing the importance of outliers. The final number of clusters identified (five) was chosen to balance clear distinct groups and ease of interpretability of results.
Table 5.4 reports the average score per function, within each of the five clusters. The scores are standardised to mean zero and variance 100. Accordingly, values above zero indicate above-average performance, while values above 100 correspond to one standard deviation above the mean.
Results show that intermediary cities are rather distinct with respect to the functions that they provide. Both particularly high and low function scores can help interpret clusters. Four out of five clusters score highest in at least one function. Conversely, Cluster 5 does not lead in any of the seven functions and is instead characterised by scoring lowest in two functions.
Based on their functions’ scores, the five clusters may be interpreted as different intermediary city types:
Cluster 1: “Knowledge Hub Intermediary City”. This city type hosts universities and other higher education institutions, and it offers the strongest business services among the five types.
Cluster 2: “Service Hub Intermediary City”. This city type offers the highest number of shopping centres and hospitals per capita among the five types, benefitting from high road transport performance. Business services are the most limited.
Cluster 3: “Housing Hub Intermediary City”. This city type provides a key residential role, with the largest share of workers commuting outside the city boundaries among the five types, supported by strong road transport performance.
Cluster 4: “Cultural Hub Intermediary City”. This city type provides the highest number of cultural amenities per capita, benefitting also from strong road transport performance.
Cluster 5: “Self-contained Employment Hub Intermediary City”. This city type offers above average business services, but weak road transport performance, reflected in the low share of workers commuting beyond city boundaries. Chapter 5.3 further highlights the self-contained nature of this type’s labour market, showing the type is further away from larger cities and has a lower share of inbound commuters.
Standardised function scores by cluster
Note: Table reports average function scores within each cluster, not weighted by population. Function scores are standardised to mean zero and variance 100. A function score of 100 should be read as a one standard deviation above the mean. Standardisation based on the sample of 173 intermediary cities from Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden included in the clustering.
Source: Analysis based on DG REGIO (2019[2]); Eurostat GISCO (2023[4]); employment-by-industry data from national sources; OECD Regions, Cities and Local Areas database http://oe.cd/geostats; Overture Maps Foundation (2024[3]); Statistical Offices of the Federation and the Länder (2024[5]).
Intermediary city types are distributed across countries rather than confined within them (Figure 5.9) (Table A.B.1 in Annex B). The clustering exercise pools intermediary cities across countries, classifying them based on function without any direct dependence on the country. The pattern that intermediary city types span across countries highlights similarities in how cities function across Europe. This is most clearly illustrated in countries with a large number of intermediary cities included in the clustering analysis, namely Germany (44), Italy (58) and Spain (52). In each of these countries, all city types are represented.
However, the distribution of city types is not even across countries. Germany has a disproportionate share of Service Hub Intermediary Cities, at almost 40%. In Italy, Housing Hubs and Self-contained Employment Hubs dominate, together making up around three-quarters of cities. The Italian case study city, Brindisi (OECD, 2026[6]), is one of three Italian Service Hub cities. Spain shows a different pattern. Half of Spanish intermediary cities are Service Hubs, while a further quarter are Housing Hubs.
The picture is more varied in countries with fewer intermediary cities. Austria’s only intermediary city, Klagenfurt (OECD, 2026[7]), is classified as a Self-contained Employment Hub. In Latvia, two of the three intermediary cities fall into this category, including the case study city Liepaja (OECD, 2026[8]). Portugal is characterised mainly by Housing Hubs, which account for five of the seven cities included. Sweden, by contrast, has only eight intermediary cities, but three are Knowledge Hubs, giving it the highest share of this type.
Intermediary cities by cluster and city size
Note: Sample of cities consists of 173 intermediary cities in Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden.
Source: Analysis based on DG REGIO (2019[2]); Eurostat GISCO (2023[4]); employment-by-industry data from national sources; OECD Regions, Cities and Local Areas database http://oe.cd/geostats; Overture Maps Foundation (2024[3]); Statistical Offices of the Federation and the Länder (2024[5]).
Overall, this experimental typology shows that similar intermediary city profiles emerge across countries with different labour market structures, industrial compositions, education systems and governance arrangements, suggesting that the typology captures broader structural patterns rather than country-specific characteristics. We view the current typology as an important first step and would welcome future work to test and refine the framework in a wider range of OECD and non-OECD contexts, including North America, Latin America and the Caribbean, thereby strengthening its international relevance and applicability.
Beyond the functions they perform, the five intermediary city types differ in their demographic, geographic, environmental, governance and economic characteristics. Comparing these characteristics helps situate each city type in a broader context and provides a fuller picture of the kinds of intermediary cities captured by the typology. Table 5.5 compares 14 indicators across city types and Figure 5.10 benchmarks gross income per capita, used as a proxy for economic performance, relative to national averages.
The comparison reveals distinct city profiles across intermediary city types: Knowledge Hubs combine high educational attainment, strong economic performance and an important governance role; Service Hubs are characterised by low income levels and limited governance roles; Housing Hubs are the most dense, with lower educational attainment, higher environmental pressures and limited governance roles; Cultural Hubs combine relatively high educational attainment with weaker economic performance; and Self-contained Employment Hubs are larger, less dense and more remote, with average economic performance.
Summary statistics for each cluster
Note: Indicators refer to 2024 or the latest available year. Standard errors in parentheses. Sample consists of 173 intermediary cities in Austria, Germany, Italy, Latvia, Portugal, Spain and Sweden. Population density is expressed as resident population per square kilometre. Net migration rate is defined as in-migration minus out-migration per 1 000 residents, including internal and international flows. Share with low education is the population share with an educational attainment classified under ISCED groups 0 to 2 in ICs relative to the national average. Distance from the nearest large city refers to the straight-line distance (in km) between an intermediary city and the nearest city with more than 1.5 million residents. The nearest large city is allowed to lie outside the country. Temperature change is the difference in air temperature in °C measured at height of 2m (in 2023 or latest available) minus the average temperature over the period 1981-2010. NO2 exposure is defined as the population-weighted average air pollutant concentration. Share of regional capitals is reported for the 170 intermediary cities excluding Latvia, as Latvia does not define regional capitals as an administrative category.
Source: Calculations based on OECD Municipal and Local Area Database http://oe.cd/geostats.
Gross income per capita relative to national average by intermediary city type (%), 2023 or latest
Note: Gross income per capita from 2023 or latest year available, expressed in percent relative to the national average. Income is averaged by cluster (unweighted aggregation). Parentheses show average gross income/capita gap in levels in 2022 USD PPP. Country sample consists of Austria, Germany, Italy, Portugal, Spain and Sweden. Gross income per capita is not available for Latvia.
Source: Calculations based on OECD Municipal and Local Area Database http://oe.cd/geostats.
Knowledge Hub Cities are the largest of the five intermediary city types, with an average population of 174 000. They have the most highly educated populations: the share of residents with low educational attainment is 13% below the national average. Their economic performance is the strongest among the five types, with gross income per capita only 1% below the national average. This combination of size, skills and income is consistent with this type’s strong knowledge and business functions, where human capital supports high value-added, knowledge-intensive activities. They play an important governance role, with one in three cities serving as a regional capital. Spatially, they combine moderate density with the highest share of urban green areas (65%). They are also least likely to be at the coast, with three in four cities situated inland.
Service Hub Cities are on the smaller size, with an average population of around 140 000. Educational attainment is similar to the national average, yet income levels lag further behind than in any other type. Gross income per capita is around 12% below the national average, corresponding to an annual gap of roughly USD 3 400 (2022 PPP) per resident. This weaker economic performance is consistent with the relatively limited presence of business functions in this city type. They play a limited governance role, with very few regional capitals (2 out of 49). Population density is moderate at 359 residents per km², and the urban green area share is the lowest among the five types (57%).
Housing Hub Cities are relatively small at around 137 500 residents. They have the lowest levels of educational attainment, with the share of low-educated residents 6% above the national average. Gross income per capita ranks in the middle of the five types. The average economic performance is consistent with the average business function of this type. Housing Hub Cities rarely serve as regional capitals (1 out of 50). Geographically, they tend to be located closer to large cities of over 1.5 million residents than the intermediary city average. 40% of them are situated on the coast. The geographic position aligns with the strong housing function characteristic of the type. They have the highest share of workers commuting outside the city and relatively strong transport performance. Coastal locations near larger cities can offer attractive living environments while providing access to external labour markets. Cities of this type record the highest level of air pollution, with NO2 exposure nearly 25% above the intermediary city average.
Cultural Hub Cities are the smallest type, with an average population of around 136 500 residents. They have relatively well-educated populations, with 7% fewer residents of low educational attainment than the national average. They have the highest net migration rate among the five types. Yet, they record gross income per capita 7% lower than the national averages, making them the second weakest economic performers among the five types. The gap suggests that available human capital is not fully translated into economic outcomes. Around 20% of these cities serve as regional capitals (3 out of 15). In environmental terms, they have experienced the largest increase in temperatures among the five types.
Self-contained Employment Hub Cities are the second largest city type with over 170 000 residents. Educational attainment is close to the national average. Their economic performance is relatively strong, with the second highest gross income per capita relative to the national average among the five types. The workforce is more local than in the other city types. The cities are characterised by the least pronounced housing function, as they feature the lowest share of workers commuting outside the city. They also attract fewer inbound commuters. Workers from outside the city account for 18% of the city workforce, 25% below the intermediary city average. They are the least densely populated with just under 260 residents per square kilometre, around 30% below the intermediary city average. Geographically, they are the most remote from larger cities of over 1.5 million residents. The average distance exceeds 200 km, and 39% lie on the coast. This combination of low density and remoteness is consistent with a weak transport function and a self-contained labour market characterising this city type.
References
[2] European Commission: Directorate-General for Regional and Urban Policy (2019), “Road transport performance in Europe – Introducing a new accessibility framework”, https://data.europa.eu/doi/10.2776/046835.
[1] Eurostat (n.d.), Statistics Explained, Glossary: Knowledge-intensive services (KIS), https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Knowledge-intensive_services_(KIS) (accessed on 4 June 2026).
[4] Eurostat GISCO (2023), Healthcare services in Europe, https://gisco-services.ec.europa.eu/pub/healthcare/ (accessed on 1 July 2024).
[6] OECD (2026), Unlocking the potential of intermediary cities for regional development in Brindisi, Italy, OECD publishing, Paris, https://doi.org/10.1787/b3a6ab2c-en.
[7] OECD (2026), Unlocking the potential of intermediary cities for regional development in Klagenfurt, Austria, OECD publishing, Paris, https://doi.org/10.1787/df0052f0-en.
[8] OECD (2026), Unlocking the potential of intermediary cities for regional development in Liepaja-Saldus, Latvia, OECD publishing, Paris, https://doi.org/10.1787/a6a418d3-en.
[9] OECD (2025), Shrinking Smartly and Sustainably: Strategies for Action, OECD Rural Studies, OECD Publishing, Paris, https://doi.org/10.1787/f91693e3-en.
[3] Overture Maps Foundation (2024), , http://overturemaps.org (accessed on 22 May 2024).
[5] Statistical Offices of the Federation and the Länder (2024), Verzeichnis der Krankenhäuser und Vorsorge- oder Rehabilitationseinrichtungen in Deutschland, https://www.statistikportal.de/sites/default/files/2026-05/krankenhausverzeichnis_24_0.xlsx (accessed on 11 March 2026).
Note