1This paper focuses on international cooperation between European cities. The intent of this cooperation is to connect cities and invite them to exchange on issues that they deal with, such as urbanism, public transport, the environment, and so on. These links form a cooperation network, and the aim of this paper is to study that network on the European scale. The main goal of this study is to test the concept of rescaling to explain that cities take advantage of cooperation to gain a new place in the hierarchy of cities and to have an increased political weight on a national and European scale. Thus, city-to-city cooperation would reveal a parallel urban hierarchy resulting from horizontal links (cooperative relationships) between cities.
2The amount of cooperation between cities has surged since the end of the Second World War (Zelinsky, 1991; Acuto, 2016). Relationships between cities have always existed, but were considerably reduced by the construction of nation-states; we are now witnessing the return of the cities (Ewen et Hebbert 2007; Le Galès 2003; Barber 2013). A new dynamism gradually arose during the second half of the twentieth century for diverse reasons, such as the creation of the European Union, the globalisation of exchanges, and the decentralisation of state competences to subnational levels. These relationships created a “city diplomacy”, that paralleled the state diplomacy (Acuto, Morissette, Tsouros, 2017; Curtis, 2014; Espineira, 2020). Currently, cities are particularly active in environmental issues, and work within many related associations and projects, such as ICLEI or C40 (Acuto et al., 2017; Emelianoff, 2007). This renewal leads to the multiplication but also the diversification of forms of cooperation between cities. The traditional form of cooperation – city twinning – is complemented by new types of cooperation, such as city associations or time-limited projects.
3In the European context, city-to-city cooperations can be divided into three types – twinning, city associations, and European programs, with respective characters (Table 1).
Table 1: Types of international cooperation between cities
4Different practices and networks stem from the richness of these cooperation types in Europe. In our paper, we will only focus on European cooperation programs and more specifically on Interreg C and Urbact programs for the 2000-2006, 2007-2013, and 2014-2020 periods. Indeed, interurban cooperation is an ultimate aim in the construction of the European Union. European institutions finance cooperation mainly through the cohesion policy, and different programs have been created for each cooperation type (transnational, cross-border, interregional, etc.) Interreg A, B, C, Urbact. This long-lasting system of cooperation has been networking European territories and especially cities for 30 years and promotes Europeanisation. Cooperation between cities is part of the Europeanisation process (Hamedinger, Wolffhardt, 2010) because, through those links, ‘good practices’ – namely in urbanism, policymaking, and the environment – circulate among local actors. Actors at a local level introduce European “ways of doing things”, which create a European identity and a feeling of participating in the construction of the European project. Although cooperation links can be considered weak and faint, they outline political relations at a local level inside community territories and help the circulation of expertise and good practice among elected representatives. This point has been addressed in the academic literature in political science and urban planning (Dühr, Nadin, 2007; Payre, Spahic, 2013).
5Studying cooperations between European cities leads to the importance of the network’s concept in academic literature. Cooperation between cities is part of a globalised world where connections and networks are essential. In a work entitled The Rise of the Network Society (1996), Manuel Castells explains that the “space of flows” has replaced the “space of places”. Networks discriminate and differentiate between spaces at every level in every component (social media, transportation, economics). The topological dimension supplants the topographical one in the organisation of space and geographical distance is less important than the existing network’s links regarding relations between places. Hence, Interreg C and Urbact programs connect partners that do not border each other. The network formed by such programs leads us to rethink the notions of distance and closeness between cities.
6Studying places through a network-based approach renews both social and urban studies. Zachary Neal explains that “while traditional social science approaches concentrate on the individual characteristics of places to explain and understand the social world, a network-based approach concentrates instead on the complex patterns of relationships between these places” (2012, p. 3). Roberta Capello adds that, “through participation in the network, cities exploit scale economies in complementary relationships and synergies in co-operative activities” (2000, p. 1945). Consequently, many researchers have studied the impact of network affiliation for the development of a specific place, and more particularly the impact for cities. Regarding European cities, among the most recent outstanding works, researchers have studied participation in innovation networks (Comin, 2009), in scientific collaborations (Maisonobe, Eckert, Grossetti, Jégou, Milard, 2016; Maisonobe, Jégou, Cabanac, 2018) or in academic partnerships in order to address metropolitan functions and urban systems in a competing world of cities. Our contribution deals with a relational approach of cities involved in European cooperation programs (Interreg and Urbact) stating that connectivity not only strengthens metropolitan functions but also gives rise to other forms of networked city systems.
7Another way of looking at intercity cooperation is to study the paradox created by the simultaneous growth of competition and cooperative links. Indeed, the advent of a liberal economy in a globalised world has led to growing competition between places (Camagni, 2002; Fujita, Krugman, Venables, 1999), which have struggled to attract more inhabitants, head offices and cultural institutions, and so on. Public resources have become rare, and states eroded the equal distribution of funds between places by calling for projects that reinforced competition between them. This rivalry has affected cities; more specifically population concentration and political, economic, cultural, and financial functions have enabled and even forced cities to be competitive in this liberal system (Sassen, 1991, 2018; Veltz, 2014). Cities implement strategies to be more visible on the international scene and gather more advantages for their development. European policies contribute to this liberal context by enhancing both competition and cooperation translated by competitiveness and cohesion in the European language. We also place our study in this approach.
8This contradictory vitality of cooperation in a context of increasing competitiveness can be explained by the concept of “coopetition” developed by researchers in management. Coopetition is a porte-manteau word defined as “a strategic and dynamic process in which economic actors jointly create value through cooperative interaction, while they simultaneously compete to capture part of that value” (Bouncken et al. 2015, p. 15). Usually applied to relations between firms and companies, this notion describes in an efficient manner relations between cities. Cooperation between cities is indeed a means for being competitive. The diversification of forms of cooperation meets the need for competitive strategies, as cooperation networks enable cities to gain more visibility at an international scale, to export their image and expertise and to be aware of new “good practices” (Jayne, Hubbard, Bell, 2011; Payre, Spahic, 2013). These networks are used as lobbies towards national or supranational organisations. They are used to represent cities' position to supranational institutions, particularly to support the will of cities to have more political power and competencies. Studying the cooperation between cities therefore circumvents the traditional one scale approach of competition between cities by promoting a multilevel and relational approach through cooperation.
9Furthermore, the aim of our paper is not only to consider cities as actors, but also as connectors (Rozenblat, Melancon, 2013) whose networks at different scales draw a new geography of European cities. We intend to go beyond a political approach in order to focus on the spatial dimension of interurban cooperation. Indeed the study of international cooperation between cities is a dynamic field of research in political science where researchers have analysed such cooperation through several approaches. The first approach focuses on the construction of a city’s diplomatic scene, in a way it parallels state diplomacy. The decentralisation of political competences enables subnational territories to forgo national control and forge international links by themselves. This penetration of cities on an international political scale complicates international relations. Diplomacy is no longer stamped by an “embedded statism” (Taylor 2000) because subnational levels are increasingly important (Acuto, 2013; Cole, Payre, 2016; Oosterlynck et al., 2018). The second approach takes intoaccount the densification and complexification of cooperation between cities. Initially, cooperation between cities was bilateral and mostly based on cultural exchanges (Joenniemi, 2017). However, as the competitive pitch between cities led to a redefinition of cooperation with new forms and goals, today’s cooperation is mostly multilateral (i.e., involves multiple actors), and its goals enter into the scope of territorial branding. The latest approach to analysing inter-urban cooperation suggests that the institutionalization of cooperative structures supports the exchanges of best practices between cities and the circulation of urban models (Huré, 2009). However, from a geographical perspective, the political approach described above does not study the spatial effects of inter-urban cooperations. Some political scientists state that cooperative relations between cities are “territorially disintegrated networks” (Heeg, Klagge, Ossenbruüge, 2003, p. 144); this appears to us as an incorrect analysis of the situation.
10In geographical studies, city-to-city cooperations as networks have not received sufficient attention. Urban networks have mostly been addressed through urban systems, highlighting the fact that they need to be documented by both flows with material dimension (such as trade and transport) and immaterial dimension (for example knowledge and innovations). Such an identification of urban networks relies on in-depth analysis whether in terms of functional completeness (Rozenblat, Melancon, 2013) or in terms of spatial extension and interconnectivity (Berroir et al., 2020). Although they display a relational approach, such papers corroborate a perspective based on hierarchy of urban systems or their proximity and discontinuity into space. A second approach rather focuses on links than on flows. It is based on a local level approach to understand the interest of cooperation between local actors. Cooperation is then conceived as a tool for learning and getting to know each other over time, based on shared values or experiences regardless of borders and territories (Baldersheim, Bucek, Swianiewicz, 2002; Zelinsky, 1991). Cooperation is studied in terms of practices such as hospitality (Jayne, Hubbard, Bell, 2013), friendship and benevolence (Clarke, 2011) or thematic concerns (Bansard, Pattberg, Widerberg, 2017; Lee, Jung, 2018). Cooperation then produces a social proximity between cities based on chosen relations. Such analyses require an intimate knowledge of the histories and modalities of twinning; however, they make comparison difficult. Thirdly, another approach is based on the rescaling process or ‘politics of scale’. It states that cities are taking advantage of cooperation to gain both a new place in the urban hierarchy and a political weight at both national and international levels (Brenner, 2003; Cox, 2009; Swyngedouw, 2000, 2004). Following a competitive and liberal approach, rescaling introduces a shift into scales from the local scale to the global one, where state level is insufficient to address globalisation challenges that cities are facing. Some other authors have shown that rescaling is a means for European cities to challenge urban hierarchies but also to bring out a new intermediate regional scale such as Baltic sea or Black sea regions where cities can consolidate their place (Boulineau, 2016, 2017; Escach, 2014; Escach, Vaudor, 2014).
11Our goal is to highlight city-to-city cooperation networks in Europe. Following some insights of the geographical literature, we state that this cooperation draws new networks based rather on links than on flows. This geography of urban cooperations relies on horizontal links not integrated in functional hierarchies but in a chosen participation to programs financed by the European Union. One could believe that such elective links are weak and unsubstantial but network theories have already emphasized the strength of weak ties regarding networks (Granovetter, 1973). We also consider these cooperation networks as a lever for cities to escape from the national level to the European one by exploiting the opportunity of European programs, acting both as actor and connector. In that sense they should fully comply with the rescaling process. In order to test this set of assertions we develop a network analysis based on selected European programs gathered in keep database.
12We propose to analyse databases that collect and collate cooperation programs between European cities. Our aim is to unpack the information available on those databases and examine the limits of this information. To study the structure of these cooperation networks, we have extracted databases from the website “https://www.keep.eu/”. Keep.eu is a website updated by an Interact project. It lists all European cooperation projects dividing them into cross-border, transnational and interregional cooperation programmes between European Union member states and between member states and neighboring countries. The database covers three periods: 2000-2006, 2007-2013 and 2014-2020. Even though the programming period went until 2020, our data collection ran until October 2019. For this paper, we have collected the data of Urbact, a specific program dedicated to urban development and, Interreg C a cooperation strand turned to interregional cooperation. Both programs aim at networking cities beyond borders and without spatial contiguity.
13
Table 2: Main figures of the keep database extraction

14A process of selecting and modifying the information was necessary after collecting the data. Indeed, the databases list every partner in each cooperation project. For the purposes of our study, we have retained partners which were cities and eliminated the other partners (subnational-level entities such as regions, companies, universities and associations) in order to better comply with our research question. As one can see in Table 2, the Interreg C programs are composed of many types of partners, while Urbact programs are mainly formed by cities.
15Once we made this selection, we transformed the databases into a flow matrix (Figure 1). In addition to this matrix, we built a database to collect information on the cities (geographic coordinates, population, level in national urban hierarchy and more).
Figure 1: From a Table to a flow matrix
16The originality of this study is to analyse European cooperation projects not as isolated groups but as a coherent set where information and relationships flow beyond the specific project. Thus, we built seven matrices, one for each cooperation program and one combining every program (Table 3).
Table 3: Networks descriptions
17This bi-modal network (partners associated to projects) was transformed into a unimodal network, with city-to-city links (see Figure 2). The aim is to study the structure of these networks and to identify the important cities. In order to do so, we analysed this data using the Gephi software for network visualisation, made maps using Qgis, and calculated descriptive statistics with R.
Figure 2: From a bimodal to a unimodal network
18The structure of the global cooperation network matches both well-known models in graph theory: the small-world model and the scale-free model. A glossary of the terms used in this paper can be found in Table 4.
-
Barabasi and Albert proposed a scale-free network theory in 1999. A scale-free network is one where degree distribution follows a power law, which means that a few nodes have many links and the majority of nodes have few links.
-
Watts and Strogatz proposed a theory of a small-world network in 1998. In such a network, most nodes are not neighbors, but two nodes can be connected through a small number of nodes. To ascertain if a network matches the small-world model, the average path length must be lower and the clustering coefficient higher than in a random graph.
Table 4: Network analysis lexicon
Degree: The degree of a node is the number of edges connected to the node. In our study, it is the number of cooperation links of a city.
Betweenness centrality: A score which reveals the number of times a node acts as a bridge along the shortest path between two other nodes.
Diameter: The longest of all the calculated shortest paths in a network.
Density: A ratio of the number of edges to the number of possible edges in a network with nodes.
Modularity: The strength of division of a network into modules (also called groups, clusters or communities). Networks with high modularity have dense connections between nodes within modules but sparse connections between nodes in different modules.
Community: A subset of nodes within the graph such that connections between the nodes are denser than connections with the rest of the network.
Clustering coefficient: The global clustering coefficient is the number of closed triplets over the total number of triplets (both open and closed).
Connected component: The maximal set of nodes such that each pair of nodes is connected by a path.
Average shortest path length: The average shortest path length is calculated by finding the shortest path between all pairs of nodes and taking the average of all the lengths of those paths.
|
19The network’s degree distribution corresponds to a power law, as can be seen in Figure 3. Hence, a few cities have many cooperation partners in the network, while the rest have a small number of cooperation links.
Figure 3: City degree distribution following a power law
20The path length of the global network is 2,989 and the clustering coefficient 0.667, matching the small-world model. Thus, cities in this cooperation network are not “far” from each other on the topological dimension, which means that any two cities in the network can reach each other through a short sequence of acquaintances. Indeed, if the city of Sevilla wants to exchange good practices with the city of Warsaw, actors from Sevilla can ask the city of Aarhus to put it in contact with Warsaw, as Aarhus cooperates with both Sevilla and Warsaw.
- 1 This test correlate the structure of this network with a network that has an ideal core-periphery s (...)
21As this network matches both models, it leads us to conclude that the network is organised into a core-periphery structure (validated by the Quadratic Assignment Procedure test1), with a small number of very central nodes which organise the flow of information inside the cooperation network. As one can see in Figure 4, some cities are very central because they have a high degree and a strong betweenness centrality. They are cut points because removing 3% of the cities with the highest degrees increases the number of connected components from 14 to 21. Thus, those central cities arrange the network and its flow.
Figure 4: Cooperation links within Urbact and Interreg C programs (2000-2019)
22To go further, we have studied each cooperation program separately, and it has led us to define two types of cooperation networks. Indeed, the indicators of Interreg C and Urbact programs are very different, as we can see in Table 5. Interreg C programs are more divided than Urbact programs. The former has many connected components, whereas the latter have only a few connected components. The flow of information and cooperation between cities is more limited in Interreg C programs. As illustrated in Table 5, the density is higher in Urbact programs. The global clustering coefficient is also higher in Urbact than in Interreg C programs. As can be seen in Table 5, cooperation programs from the new programming period follow this trend. The actual data, gathered in October 2019, confirms this result and corroborates the difference between Urbact and Interreg C programs, with a dense and unified Urbact community, and a divided Interreg network. These two factors (connected components and density) lead to a great difference of modularity class between the programs. Urbact cities evolve in a unique system, whereas Interreg C cities cooperate in several divided groups. This structure difference is interesting because it leads to contrasting types of cooperation. In Urbact, the cooperation network goes beyond specific cooperation projects: Urbact cities seem to be a united community wherein exchanges and expertise overtake projects, and information circulates easily throughout the entire network. In contrast, the Interreg C network seems more centered on specific projects, and the information flow between cities is reduced.
23However, this measure may be criticised because Interreg C, as we have seen, is made up of different types of partners (not just cities). In our study, we only kept cities. It is a limit of this study, but a limit that we undertook. We only study city-to-city relationships because, as it has been highlighted in the literature review, cities are both connectors and actors, being at the same time geographical entities and political actors. Their political influence is growing, especially on the European scale. The links between those cities are a fundamental tool used to claim more inclusion. Therefore, this study analyses the means that are offered by the EU to acknowledge those links and those demands. As we only focused on city cooperation links, we did not study other nodes (universities, companies, associations…).
Table 5: Network indicators by type of programs
24As the nodes are cities, it is interesting to go beyond the topological analysis and to put cities back in a spatial dimension. As we can see on the map (Figure 5), some important cities stand out. It is interesting to see that central cities in the topological space of the cooperation network are not necessarily central cities in the topographical space or important cities from a political, demographic and functional perspective. Nicolas Escach studies the example of Riga and Vilnius in his PhD thesis (Escach 2014), explaining that cities around the Baltic Sea use cooperation to gain some political influence and centrality in the EU city hierarchy. Escach uses the concept of rescaling to emphasise the idea what those cities’ political- and spatial-level changes owe to cooperation. This statement is supported by the fact that the most important European cities, such as capital cities, large cities playing an international role or even the two global cities Paris and London (Sassen 1991), are not overrepresented in the European cooperation network. The GAWC research group, in studying the global city network, explains that there is a pre-established network between “global cities” (Taylor, 2004). The cooperation network can yet be seen as a way to create alternative links and relations for less important cities to become more significant and more influent on the European scale. The impact of the concept of rescaling is legitimated through this analysis. The large number of port towns and their significant weight in the network (Dublin, Thessaloniki, Riga, Napoli…) is also noteworthy, underlining the specificities of those cities with respect to their international openness.
Figure 5: Centrality of European cities involved in cooperation projects (2000-2019)
25To support the rescaling theory, the following figures show that cities which are the best integrated in this cooperation network are both regional capitals and cities between 500,000 and 1 million inhabitants. Figure 6 shows that some regional capitals (Belfast,Torino, Francfort, Sevilla, etc..) have as many partners as national capital cities (Budapest, Sofia, Vilnius, Riga). Figure 7 shows that cities having the largest number of cooperation partners are not the most populated ones, but cities below 1 million inhabitants.
Figure 6: Number of cooperation partners according to the administrative status of cities
Figure 7: Number of partners according to the population
26After analysing the places of cities, we can now look at the web of cooperations connecting those cities. Most of the ties between cities are unique and last for one programming period. Indeed, 95% of the ties between cities occur for only one programming period, and 90% of the ties are constituted by one cooperation project in common. Therefore, most of the cooperation links between cities are fleeting. The evaluation committee on Interreg II points out this transitory nature of cooperation links, because of the unsatisfying results that come with it (Boulineau 2016).
27Nonetheless, some cooperation links seem stronger and more intense. We selected ties that last for at least two programming periods and involve at least two cooperation projects in common. The result is a corpus of 144 cities and 636 cooperation links (Figure 8). Although it displays a smaller number of edges, this network still looks dense and is composed of many cities. Out of the initial city corpus, 21% is represented on this map, even if only 5.6% of the links are represented. These stronger relations reveal an important implication of cities in European projects, even if each bilateral link cannot be maintained. The spatial analysis shows that the distance between two cities is not an important aspect to explain a lasting cooperative relationship (see Napoli-Vilnius, Glasgow-Torino, Belfast-Lyon…). This map raises questions about the underlying factors which lead to a persistent cooperative relationship and why cooperation links seem to last so rarely. The transitory nature of city-to-city relations is indeed a crucial issue. It has been deepened by fieldworks interviews realized alongside the database analysis but it was not possible to mention them in this single paper.
Figure 8: Some intense and lasting links between cities
28Lastly, Figure 9 displays on which thematic interurban cooperation is based. As we can see, the most common thematic is “urban development”, which is logical because cooperation takes place between cities. “Economic development” is the second most prominent thematic, it confirms that cooperation between cities has an economic objective and is no longer centered on cultural and symbolical exchanges as underlined by the literature. The same comment can be made about the place of “innovation” as a cooperation thematic. The significant part of “planning” follows the logic of cooperation as a way to exchange technical expertise between elected representatives and urban employees. “Governance and cooperation” also have a great weight, which reinforces that exchanges and cooperation are a leading topic in the European governance driven by the concept of subsidiarity. The analysis on the evolution of each thematic over time must be carefully driven, because the last programming period is not complete and the proportions can still be modified after 2020. However, we can define some interesting elements. The first major trend is the increased proportion of “green development” projects over the three programming periods. Even though the last programming period is not yet over, the number of projects centered on green development exceeds that for the last two periods. This increasing prominence of green development is patent because of the growing stakes regarding the environment and climate change. The other noteworthy evolution is that of agriculture, characterised by significant growth over the last programming period. This development is also explained by expanding concerns about ecology, because the majority of the projects are related to green and sustainable agriculture, as well as the issue of urban farming.
Figure 9: Subjects of European cooperation between cities by programming period
29Figure 10 shows that the cooperation differences between Interreg C and Urbact programs are also observed in projects’ objectives. Indeed, the cooperation topics of Interreg C contrast with those of Urbact. Interreg C subjects are shared more regularly than in Urbact, because it has a larger panel of topics. The topics which stand out in Interreg C programs are planning, economic development, innovation and green development. Urbact’s projects are focused on two themes: urban development as well as governance and cooperation. This interesting result supports the idea that Interreg C projects are centered on a specific project dedicated to one thematic. Thus, cooperation themes are diversified because they match the concrete need of every partner’s group. On the other hand, Urbact projects are mainly devoted to cooperation between cities and are less about a specific thematic. This result corroborates the network structure analysis and supports the understanding of Interreg C and Urbact as two distinct types of cooperation, used differently by cities depending on their needs.
Figure 10: Interreg C and Urbact : different cooperation focus ?
30Keep databases are complete and globally well-structured. There are some misprints in the tables, but they are a reliable source after slight corrections. However, some qualitative information is often missing: budget, expected results and achievements. The data is more accurate for the periods 2014-2020 and 2007-2013 than for 2000-2006, as well as more accurate for Interreg C than for Urbact programs. The quality of the analysis is still not the same for each database. Budget analysis would be relevant, with the aim to compare resources allocated depending on the thematic or project partner. The amount of information is very significant despite the mentioned limits, and it can be used in many ways (not merely in a geographical study), revealing for example what the most important theme is how long a project lasts. The potential of these databases is broad, and more studies should be based on them.
31Another limit is the lack of one year for the last programming period. Some interesting results for this period cannot be validated yet because of the missing data. Two results must be noted, the first concerns the declining participation of Western cities in Interreg C, and particularly the French ones: there were 36 French cities in 2000-2006, 18 in 2007-2013 and, for now, only 5 cities for the 2014-2020 period. The second noteworthy result is the decreasing participation of cities among both Interreg C and Urbact programs, as can be seen in Table 2, which leads us to think that those projects include new types of partners (businesses, associations, universities). The increasing presence of new actors within cooperation projects is described in the literature on political science (Rozier 2017; Huré 2013). Unfortunately, only complete data will allow validation of these assumptions.
32In this paper, we have analysed some databases extracted from the website “keep.eu” which lists different European cooperation programs. These programs are based on the exchange of information and good practices between partners. After analysis, we have shown that the European cooperation network is a small-world and scale-free network, structured on a core-periphery model. Nonetheless, two types of networks can be identified: Urbact and Interreg C. Their different structures lead to a divergent circulation of information. This difference is also visible in the weight of different topics in each program. Thus, very central cities control the information flow, but these are not necessarily large cities on the European scale. The concept of rescaling is highly functional because it demonstrates how cities gain political influence through the strategic repositioning.
33To conclude this paper, we can raise the question of the impact of those cooperation links on the urban hierarchy. Does the participation to this cooperation network jostle the European urban system? Cities, through European cooperations, have access to funds, new know-how and different experiences that take them out of the domestic context. In an interview, the director of Urbact explained that the city of Alba Iulia in Romania, thanks to its participation to Urbact projects during 20 years, gained over 300 million euros. According to him, it allowed financing of its action plans with European funds in order to transform the image of the city and climb the ladder of urban hierarchy. But it is a rare example, because, as it has been said, the involvement in this cooperation network is fleeting. The irregularity of these cooperation links reduces the durability of the impacts on the European urban system. Furthermore, those cooperation programs are recent and we would be able to see the effects in the long term.
34More generally, this kind of analysis does not aim to prove that cooperation links between cities jostle the urban hierarchy as it is usually presented in the scientific literature – that is to say mainly structured with the predominant core/periphery model and more recently driven by the paradigm of global cities and of metropolisation. Our analysis aims to shed light on other approaches of urban “systems”, based on voluntary relationships between cities. We wrote system with quotation marks because we have demonstrated that relationships established throughout European programs are somewhat labile, and do not constitute a solid and stable set of links which could be defined as a system. Thus, studies about cooperation links between cities provide meaningful results because it defines urban hierarchy not on the basis of demographic or economical attributes of cities but on their capacity and will to establish (or maintain) relationships with other cities. Such a relational hierarchy of cities is necessarily more labile because it depends on the political attitude and position of a few number of actors who can decide suddenly to stop or to switch the cooperation policy, and who act during a limited time (whereas a classification based on demographic attributes is more stable). This is why the temporality of cooperation links showed in this article has to be considered carefully. The leading position of some cities as central nodes and on the opposite the decline of others from one period to another cannot be analysed without the results of field surveys concerning the political situation and the system of actors.
35Furthermore, this relational hierarchy has its own geography. We have shown that the network of cities having the largest number of partners (Figure 3) is very often composed by medium sized towns (Figure 7) standing on the periphery on the EU (Belfast, Sevilla, Riga, Warsaw, etc.) and that on the contrary, largest cities and the urban core of Europe, the so-called European backbone are only slightly visible on the map of cooperation (with the exception of the Netherlands, and Torino). To go further, cities having the highest betweenness centrality are clearly far from this core (Vilnius, Budapest, Sofia, Athens, Plovdiv). Therefore, this result confirms that the topographical periphery is not an impediment to get a key position in a relational hierarchy, quite the opposite. This original geography is even reinforced when looking only at the most intense and lasting links between cities which are mostly located in countries like Poland, Romania, Italy, and Greece (Figure 5). In central and highly urbanised countries such as the United Kingdom, Germany or France, the most involved cities are peripheral. The rescaling effect is therefore attested: cities use cooperation links not to “upgrade” their position in the hierarchy of cities but to get around the obstacle of their peripheral position and location in Europe in search of economic, political and information resources. This result suggests that cooperation plays a greater role than competition as an engine of the implication of cities in European programs. But again, this needs to be confirmed by in-depth and qualitative analysis.
36Dataset can be found on www.keep.eu.