The article presents research findings of the project “Transnational Production Spaces” (2016-2019), funded by the Deutsche Forschungsgemeinschaft (DFG) under grant agreement No MI 1893/2-1.
1The structure and development of even the most remote or inconspicuous urban spaces are increasingly subject to global economic dynamics. Distant and disparate urban settings are linked by transnational production networks. In our research, we have traced the ways in which dense clothing production clusters in Istanbul’s inner city, large textile treatment plants in the metropole’s hinterland, dispersed sewing operations in re-used factory buildings in rural Bulgaria, and huge new industry parks near Ethiopia’s capital Addis Ababa are related to each other and to many other locations across the globe. All these places perform specific functions in transnational production networks of the clothing industry catering predominantly to West European markets. Although globalized commodity production and distribution decisively drive urban development and shape the built environment in these and multiple other places, they have been underrepresented in urban research so far, while the global movements of people, financial capital, information and services have spurred a large amount of research on “transnational urban spaces”, “the network society” or “global and world cities”, to name only some of the debates. However, contemporary urbanization at the diverse sites of industrial production can only be understood in relation to the globally dispersed structures of production and consumption. In turn, transnational production networks and commodity flows are formed and articulated by specific local constellations and material assemblages on the ground – architectures, urban spaces and infrastructures.
2In order to grasp this transnational character of urban space and development in places of industrial production, we argue that a relational perspective on multi-local economic processes as provided by commodity chain approaches can be usefully employed in urban research. Specifically, the Global Production Networks (GPN) framework developed in economic geography offers an as yet untapped analytical potential with regard to urban spaces, and the built environment in particular. Integrating the disciplinary perspective and methodology of the GPN approach with architectural and urban research allows a multi-scalar analysis of the spatiality of transnational production relations as well as their interplay with the built environment. Our conceptual considerations along these lines are laid out in this article, departing from a closer examination of the spatial concepts formulated in the GPN approach – such as territorial embeddedness and strategic coupling. Despite the spatial dimension of these concepts, for the most part production locations are treated merely as nodes within networks in GPN literature. A critical need to unpack these nodes and to bring the topological and territorial dimension of global production networks in “productive tension” has been identified by a number of scholars (Brown et al. 2010, Kelly 2013, Kleibert and Horner 2018, Leslie and Reimer 1999, Phelps 2017). In response to these claims we propose “grounding” the study of global production networks by paying close attention to architectures and material infrastructures, urban development and planning in specific places, and by examining their roles in the spatial (re-) configuration of production networks. In turn, this also allows for “globalizing” urban research through the relational and multi-scalar lens provided by the GPN approach: Multi-local research conceived along production networks contributes in a ground-breaking way to recent debates on transnational urban research (Krätke, Wildner, and Lanz 2012, Parnreiter 2012), planetary urbanization (Brenner and Schmid 2015) and relational approaches to urban comparison across the North-South divide (Robinson 2011, Ward 2010). This article conceptually and empirically explores the benefits of integrating urban research and the GPN approach for both of these research fields.
3We operationalize our conceptual considerations in a research framework that serves to study the interplay between globalized industrial production and urban spaces: By assessing the functions of different places in transnational production networks, on the one hand, and by studying the physical urban spaces on different scales and the way they are planned, on the other hand, our analysis captures the dynamic interrelatedness of economic processes and urban structures. Combining the perspective of large-scale economic systems with a fine-grain analysis of built structures and urban development may seem ambitious for a single paper, but it is precisely through this integration of different scales and methods of analysis that knowledge of the urban effects of production networks as well as the role of physical structures within these dynamic networks can be advanced. Hence, in this contribution, we lay out an analytical framework and discuss key findings of its application in a multi-local case study of clothing manufacturing locations in the Istanbul metropolitan region in Turkey, the South Bulgarian province Kardzhali and the periphery of Ethiopia’s capital Addis Ababa. We trace how physical urban changes within these places are related to the transnational supply chains and successive relocation processes among them (and other places), and thus aim to explore the transnational constitution of the respective urban spaces and the “mutual transformation” (Coe and Yeung 2015: 14) of the networks and places in their fine-grain and diversity.
4The sites of our multi-local case study were chosen with regard to successive location changes of clothing production for European markets in Turkey (Fig. 1). The country has been increasingly integrated in global production networks of the clothing industry since its market liberalization in the 1980s. Initially a sourcing region for cheap and simple manufacturing, it later became a location for full-package supply (including fashion design and own brand manufacturing), higher quality production and fast fashion (Tokatli and Kızılgün 2009, Neidik and Gereffi 2006). This “upgrading” of many Turkish suppliers involved a gradual relocation of labour-intensive production steps from the urbanized regions in Eastern Turkey (most notably Istanbul) to peripheral areas within the country, as well as to other countries in Eastern Europe, the MENA region, Central Asia, and Sub-Saharan Africa. Following these trends in a diachronic perspective, we focussed on two manufacturing locations in the Istanbul region established during the clothing production boom in the 1980s and 1990s, and two destinations of subsequent production offshoring by Turkish clothing firms: the South Bulgarian province Kardzhali, with a state socialist history of clothing manufacturing, and the fringes of Ethiopia’s capital Addis Ababa, a recent vanguard of globalized large-scale clothing production. In a synchronic perspective, the four case study sites emerge as parts of the same global production networks, as they are connected through direct firm relations and commodity flows, catering to the same global lead firms as customers and to common end markets (Fig. 2).
Fig. 1. Case study locations: Production relocations of Turkish clothing producers.
Map: Anke Hagemann
Fig. 2. Number of H&M suppliers by production country, 2017.
Map: Anke Hagemann, Ilkim Er. Source: H&M Group: Supplier list 2017.
5The relational comparison of the four locations reveals the site-specific, yet interdependent mutual transformation of global production networks and urban space, giving rise to most different spatial configurations such as dense industry clusters, dispersed production niches or clearly defined enclaves for export processing. But not only the spatial characteristics of the production sites show marked differences, also the intensity and scope of transnational influence vary due to different local trajectories and actor constellations. While it may seem obvious that globalized production has considerable, yet variegated impacts on urban spaces, our findings most prominently underscore the active role of the built environment and urban planning in the mutual transformation of network and space, and thus in shaping the geography of globalized production. This becomes particularly evident when it comes to the dynamic relocation of production capacities from one place to the other and to place-specific adaptations maintaining and stabilizing network integration.
6Thus, the article proceeds in three steps: In the first section, we discuss the conceptual base, existent interfaces in literature and potential benefits of integrating both fields. In the second section, we chart our analytical framework, and discuss the key findings of its application in a multi-local case study in the third section. Building on this empirical evidence, the conclusion sums up what can be gained by globalizing urban research through applying a multi-local GPN perspective and by grounding GPN research through recognizing the significance of physical structures, urban development and planning for global production networks.
7Urban researchers in the social science disciplines have been adopting a global and relational perspective and addressed the urban dimension of transnational economic networks at least since the 1980s. Especially the study of worldwide networks among global and world cities has been pushed forward in a systematic way (Friedmann 1986, Sassen 1991, Taylor 2004). However, this strand of research has so far been operating mainly on a macro-scale and concentrating on advanced producer services and flows of information, finance or highly skilled labour. Thus, only a limited range of cities was focused on, predominantly in the Global North (Robinson 2002). In contrast, research on transnational (urban) spaces has invoked more diversity in terms of places and scales, for example by studying cross-border trade relationships and migration “from below” (Smith 2001, Crang, Dwyer, and Jackson 2003, Eder and Öz 2010, Wildner 2012). While both approaches trace global economic relations from multi-local viewpoints, they have so far largely neglected empirical examination of global commodity production and circulation. Nevertheless, some significant conceptual advances were made to integrate a global commodity chain perspective. In a first comprehensive debate, it was suggested to explore the advanced producer service nexus as a link between commodity chains and world city networks (Brown et al. 2010). But few empirical studies engaged with this field beyond the macro-level (Parnreiter 2010, Vind and Fold 2010). In this context, Christof Parnreiter pointed to the significance of the built environment for constituting the uneven geographies of commodity chains and global cities and explored the transnational character of office buildings for globalized producer service firms (Parnreiter 2011, 2012). In the same vein, another “call for the integration or combination of global value chain analysis and global city research” (Krätke, Wildner, and Lanz 2012: 16–17) was brought up in the research field of transnational urban spaces. Krätke, Wildner, and Lanz emphasize the widely neglected importance of commodity production for global economic links between cities, and draft a multi-scalar framework for the analysis of transnational urban spaces, considering social as well as physical space.
8Studying transnational commodity flows between very different kinds of urban settings also offers a productive methodological lens for a more experimental, relational comparative urban research, demanded by several urban scholars (Robinson 2011, Ward 2010, McFarlane 2010) also in order to break with inherited divides such as “developed” and “underdeveloped” contexts. According to Jennifer Robinson, future comparative research could be conceptually based on “[s]ocial, economic and cultural flows of all kinds” (Robinson 2011: 8) between cities, and focus on the “different ends of the connections” (Robinson 2011: 15), such as headquarters and production centers of globally active firms. As a result, territorial perspectives on urban spaces would be complemented by a more relational spatial thinking. Commodity flows are also central to Neil Brenner’s and Christian Schmid’s theorization of “planetary urbanization”. They point to “the flexibilization of production processes and the generalization of global production networks” (Brenner and Schmid 2015: 172) as one of the key facilitating dynamics of “extended urbanization” transforming remote places and hinterlands into operational landscapes of “new export processing zones, global sweatshop regions, back office locations, data processing facilities and intermodal logistics terminals” (Brenner and Schmid 2015: 152).
9Urban research within the disciplines of architecture and urbanism equally scarcely addresses industrial locations as part of globalized production systems. In earlier contributions, we have presented first advances to connect analyses of physical urban spaces and architectures with a commodity chain perspective (Hagemann 2015, Hagemann and Beyer 2019, Beyer and Hagemann 2018) which the present article builds on. Recently, several instructive contributions by architecture scholars have addressed the circulation of commodities and the urban infrastructures of logistics with a specific focus on their interplay with the built environment and the realms of architecture and urban design (Easterling 2014, LeCavalier 2016, Lyster 2016, Hein 2018, Topalovic, Knüsel, and Jäggi 2013, Footprint 2018). Albeit without reference to global commodity chain approaches, these studies explore different sections, material and organizational aspects of global production systems in a multi-scalar and multi-local way, and some use mappings as a highly productive analytical tool to capture the structural characteristics and physical transformation of the built environment in relation to global processes and movements.
10These interrelated strands of urban research provide a fertile ground and promising entry points for the integration of a global commodity chains perspective. Indeed, this integration seems to arise as a direct consequence, next step, or exemplary realization of prominent research claims. Taking up these cues, our research approaches the relational character of places of global production as well as their physical materiality (and the planning processes shaping it) in order to achieve a synthesis of both perspectives, to the effect of building an understanding of the mutual transformation of global production networks and urban spaces.
11In order to empirically study the increasingly fragmented organization and spatial dispersion of industrial production (and service provision) across the globe, the analytical framework of Global Production Networks (GPN) was developed within the field of economic geography after 2000 (Henderson et al. 2002, Dicken et al. 2001, Coe et al. 2004, Coe, Dicken, and Hess 2008). It strongly builds on the Global Commodity Chains (GCC) approach in economic sociology (Gereffi, Korzeniewicz, and Korzeniewicz 1994), which relates back to the concept of the “commodity chain” in World Systems Theory (Hopkins and Wallerstein 1986). The Global Value Chains (GVC) concept evolved as another strand of research from these roots in the early 2000s (Humphrey and Schmitz 2001). Although the three frameworks have developed different foci, they share the aim to better understand the world economy through studying the complex relationships and strategies of firms within global production chains or networks of specific products or sectors. Most empirical research guided by these frameworks addresses the organization, power structures and distribution of profits in global production relations in order to explore possibilities for regional economic development. As we shall argue in the following, the GPN approach provides the most advanced spatial conceptualization of the respective economic processes, and thus lends itself most readily to integration into urban research. It emphasizes the complex network character of production relations in contrast to the linearity of the chain metaphor, the multi-scalarity of global production networks, and the “embeddedness” of all economic activities within the socio-spatial, historical and institutional contexts of specific regions. Moreover, GPN analysis includes additional actors beyond the firm, such as state institutions, international organizations, trade unions, and civil society (Henderson et al. 2002). The more recent concept of “GPN 2.0” seeks to update this framework through a more thorough conceptualization with the aim of contributing “to explanations of patterns of uneven territorial development in the global economy” (Coe and Yeung 2015: 22, emphasis in the original). A further critical revision of the GCC and GPN frameworks, the “Disarticulations Perspective”, aims to shift focus from the positivist notion of network integration dominating much of the literature to a more comprehensive perception of system-inherent processes of downgrading, disinvestment and disconnection from global production networks (Bair and Werner 2011a, Werner 2016b, McGrath 2017).
12Early Global Commodity Chain research expressed interest in the geographical locations where a commodity chain “touches down” (Appelbaum and Gereffi 1994: 43). In Gereffi’s initial GCC concept, “territoriality”, understood as the “spatial dispersion or concentration of production and distribution networks” (Gereffi 1994: 97), was determined as one out of three main dimensions of global commodity chains alongside the “input-output structure” and the “governance structure”. However, in GCC literature governance and upgrading became predominant issues and the territorial dimension was largely neglected (Leslie and Reimer 1999, Brown et al. 2010, Coe and Yeung 2015). Studying global networks of firms generally supports a topological conception of space, in which specific actors and their locations appear predominantly as nodes (Gereffi, Korzeniewicz, and Korzeniewicz 1994, Leslie and Reimer 1999, Kleibert and Horner 2018). Thus, Dicken et al. perceptively highlighted “the risk of losing sight altogether of profound geographical variations across localities and regions” (Dicken et al. 2001: 96), while others more recently diagnosed the susceptibility of GCC/GVC/GPN approaches to “an underappreciation of the territorial dimension” (Kleibert and Horner 2018: 226).
13The GPN framework, however, initially intended to balance the aspects of territory and network. Aiming to bring together “the analytical strength of both commodity chain analyses and theories of regional development” (Kelly 2013: 84), the sub-national region figures as the “basic geographical building block” of analysis (Coe and Yeung 2015: 18). Emphasizing the territorial embeddedness of global production networks, GPN scholars consider networks and places as mutually related:
„In order to understand the dynamics of development in a given place, then, we must comprehend how places are being transformed by flows of capital, labour, knowledge, power etc. and how, at the same time, places (or more specifically their institutional and social fabrics) are transforming those flows as they locate in place-specific domains.“
(Henderson et al. 2002: 438)
14In the same vein, Coe and Yeung’s most recent comprehensive theorization still posits the “analytical necessity to unpack the mutual transformation of both the firms/network and the places in which they are embedded” (Coe and Yeung 2015: 14, emphasis in the original). Place – or territory/geography, terms often interchangeably used in GCC/GPN literature when referring to geographically defined space as opposed to the topological and unbound space of the network – is seen to have agency within global production networks:
„Geography, however, is not just the physical surface on which these value activities are organized and performed. Rather, geography can be an active space that shapes the territorial constitution and configuration of these network activities.“
(Coe and Yeung 2015: 67–68)
15For example, territory is discussed in GPN literature as an active agent when it comes to “strategic coupling”, i.e. the insertion of regions into global production networks through “intentional and active intervention from both sides” (Coe and Yeung 2015: 20). This process is seen to match “strategic needs” of lead firms and the region’s “assets”, including economic structures, locational advantages and physical infrastructures, actively promoted, supported or established by regional or national policies and institutions (Coe et al. 2004). However, the physical preconditions which can be crucial for network inclusion are hardly discussed in the literature. While early conceptual papers have acknowledged the possibility to include also non-human actors into GPN analysis with reference to Actor-Network Theory (Henderson et al. 2002), later contributions explicitly dismissed this idea (Coe and Yeung 2015).
16Accordingly, the GPN approach does not consider the territories that global production networks are embedded in in their physical structure, but primarily with regard to “their institutional and social fabrics” (Henderson et al. 2002: 438). Hence, GPN approaches have been criticized by Philip Kelly for their tendency “to subordinate the completeness of places by focusing on their role in firms’ production networks” (Kelly 2013: 82). He demands to complement GPN analyses by place-based investigations on broader processes which “a GPN might not see” (Kelly 2013: 87), such as land-use change, livelihoods, and socio-spatial polarization – highly relevant issues for assessing regional development. Kelly suggests, “instead of resolving the dialectic between place and network in favour of networks, a productive tension between two spatial imaginaries should be maintained” (Kelly 2013: 82). Nicholas Phelps likewise argues for maintaining this fundamental dialectic between “differentiation and equalization, mobility and fixity, space and place, agglomerations and networks” (Phelps 2017: 9) in economic geography. He identifies major geographical formations in the current global economy – networks, enclaves, agglomerations, arenas – which combine topological and territorial aspects in different ways. Also Coe and Yeung (2015) identify a number of typical spatial formations at GPN nodes that correspond with specific modes of network integration, calling them “coupling types”. These include places such as innovation hubs, global cities, logistics hubs, or assembly platforms. But while Phelps detects these formations also on an urban scale, Coe and Yeung do not consider their spatial form and internal structure at all. We argue that an inquiry into such spatial figures requires comprehensive empirical groundwork on physical spaces including urban territories and architectures in order to understand their development and performance as nodes within and between global networks and local contexts. Accordingly, this paper explores how the spatiality of global production networks, as well as their interplay with the built environment, can be productively analyzed employing the disciplinary perspective and methodology of architectural and urban research.
17Kleibert and Horner (2018) similarly demand more attention for the territorial impacts of global production networks on a smaller scale. They particularly point to the socio-spatial polarization within regions and cities and thus call for an “opening-up of the ‘black box’ of the region”, which has remained the smallest scale of analysis in GPN research to date. While in most contributions to the GPN literature, production locations are treated merely as nodes within networks, several authors started to unpack specific nodes in more detailed empirical studies of the places where global production networks touch down (Kelly 2013, Kleibert 2015, Phelps, Atienza and Arias 2015, see also Pickles et al. 2016). Also, the Disarticulations Perspective emphasizes the specificity of place and its interdependency with global production networks, while at the same time highlighting the role of uneven spatial development as a precondition as well as a result of global production networks (Werner 2016b). Instead of concentrating on worldwide, multi-local firm networks, Bair and Werner study individual places within global production networks over longer periods of time, which includes the network inclusion and also exclusion of these places and the particular local circumstances, “politics of place” and historical trajectories (Bair and Werner 2011b, Werner 2016a). In response to these claims, we seek to further “unpack” network nodes also in medium and long term perspective by investigating building biographies and urban transformation processes. This allows us to trace not only the creation and functional differentiation of production sites, but also their discontinuation with the related urban effects of disinvestment, industrial decline and disuse in interplay with the dynamics of network (dis)integration.
18To sum up, the research on global commodity chains and production networks in economic geography offers very relevant conceptual connecting points and empirical approaches for a more fine-grain analysis of the spatial and physical features of network nodes. A move towards integration with urban research can especially build on the dialectic of network and place/territory and the idea of their mutual transformation that is present in most of the concepts and studies. But despite the promising approximations between both fields in existing research, there is still a need for conceptual elaboration and methodological strategies for such an endeavor, as well as for broader and deeper empirical work. The next section charts an analytical framework for researching the interplay of global production networks and urban spaces as a step towards this end, followed by its empirical application to a multi-local case study.
19As a multi-local case study integrating urban research with a commodity chain perspective, we selected four sites within overlapping global production networks of the clothing industry. These sites are connected through direct firm relations and the dynamic shift of production capacities, as well as through common end markets and global lead firms as customers. Within the selected sites, we focused on different types of manufacturing facilities located in contrasting urban places: small firms clustering in Istanbul’s dense inner-city district Merter, large industry premises sprawling around the city of Çorlu in the western periphery of the metropolis, medium-sized sewing factories in small-town settings within the rural region of Kardzhali, and gigantic greenfield industry zones in and around the growing capital city Addis Ababa.
20Our research contributes to building an understanding of how the physical structure and transformation of urban spaces in these four locations, as well as urban planning, articulate and correspond with the commodity flows, economic relations, and spatial dynamics of the global production networks they are part of. In other words, we explore the co-constitution and co-production of both, i.e. the mutual transformation of these urban spaces and global production networks: the impacts of globalized manufacturing on the built environment, and the effects of urban structures and transformations on the geography of global production networks – most apparent when it comes to production relocation processes. Moreover, through identifying the translocal forces shaping the built environment, we can grasp the transnational constitution of urban manufacturing locations.
21Combining the perspectives of the network and of urban space, we propose an analytical framework with a twofold approach (Fig. 3). The synthesis of these two perspectives allows us to establish interrelations between the two dimensions regarding their mutual transformation and the transnational constitution of urban spaces. This analytical matrix facilitates a productive combination of relational and physical readings of the spaces of globalized production, while addressing the multi-scalarity of translocal economic networks and physical urban spaces.
Fig. 3. Research framework integrating the perspectives of network and urban space.
Diagram: Anke Hagemann.
22To establish a network perspective, firstly, the connectedness and role of a specific location and of particular manufacturing firms within global production networks needs to be taken into account. In this regard, our case study design also aims to determine the dynamic interrelations between the study locations. At the outset, this involves a characterization of firms and the specific production activities they perform (including firm size, organization, ownership, location structure, development and strategy, as well as the product type, production processes, input and output). It includes tracing patterns of commodity circulation and firm relations also with regard to the (changing) function of firms in these network constellations. We seek to evaluate the embeddedness of firms and their production activity within local economic and institutional structures as well as its local economic impacts. On another level, a qualitative assessment of the places’ overall (or aggregate) role within production networks of a particular industry and the change of these roles in the dynamic processes of coupling and decoupling in the spatial restructuring of global production networks becomes relevant. As mentioned, Coe and Yeung (2015) offer a range of “coupling types” such as industry cluster, export processing zone, or wholesale hub to describe these roles. We seek to identify the roles of multi-scalar institutions, regulatory frameworks and diverse transnational and local actors within the “strategic” or conflictual coupling processes and to build an understanding of the major local “assets” coming into play, such as economic structures, labour, social reproduction, physical spaces and infrastructures.
23Examining production locations from the viewpoint of urban space, on the other hand, encompasses spatial analysis of manufacturing sites and their urban context, using analytic maps as an essential tool. This allows us to characterize spaces of globalized production and to trace their transformation on multiple scales from architecture to urban region. On the scale of architecture, analysis captures the morphology of the production facility, including building type, volume, inner structure and outer appearance. On a neighbourhood scale, we analyse the immediate urban context, including building structures (types, density and grain), land use patterns and major transport infrastructures. In the larger urban or regional context, we study the location of the respective production facilities and districts in relation to built up areas, relevant industrial sites, major transport infrastructures and significant topographical features. Studying these aspects over a certain time period can reveal major spatial transformations in interplay with production network dynamics, including the spatial expansion or withdrawal of production, and the capacity of physical spaces to accommodate production activity and its change. Of course, urban planning strategies and governance are highly significant in the urban perspective. Hence, the role of planning actors and their constellations, particularly the role of the state and of transnational actors associated with the production networks, as well as the political scale of planning decisions, are also significant aspects of analysis.
24Ultimately, our framework provides a synthesis of both perspectives and an exploration of their interplay, in order to capture the complex mutual transformation of global production networks and urban spaces. It evaluates both the impact of globalized production activity and its dynamics on urban spaces, and the impacts of urban change and planning on the (trans)formation of global production networks and their geographical structure. Moreover, an investigation from both perspectives is crucial to grasp transnational aspects of the constitution of urban spaces. This includes their configuration as “anchoring places” (Krätke, Wildner, and Lanz 2012: 10) for the global networks of processing and circulating goods (through physical infrastructure provisioning), their inhabitation by transnational actors (present as staff, customers, auditors, investors or owners, architects or construction engineers), and their aspiration to international standards (in production technology, building codes, environmental or social compliance). This also allows a comparison of the variations in intensity and scope of transnational influence, particularly with regard to novel actor alliances and their impacts on strategic coupling and planning governance.
25Seen from the double perspective of global production networks and urban space, each site of our case study exhibits specific economic trajectories and spatial structures. In the following four vignettes, we summarize key findings about each location in view of the framework’s categories to provide the groundwork for a relational comparison. In conclusion, we highlight interrelations and interdependencies correlating with transnational production arrangements by tracing the relocation dynamics through the four sites.
26Merter is a dense inner-city production and wholesale district for textile and clothing located in the Western part of Istanbul, just outside the ancient city walls. Focusing on Merter through a production network lens, we see that the firms in this area are mostly integrated into two types of global production networks: On one hand, there is a strong presence of headquarters of full-package suppliers and some of their specialized subcontractors, suppliers and service providers (such as embroidery workshops, printshops, accessories wholesalers) producing clothes for West European fashion firms and markets. On the other, there are numerous wholesale shops offering clothes especially for small traders from Eastern Europe, Russia, Central Asia and Arabic countries (often referred to as “suitcase trade”, see Piart 2012). Both rely on supply networks outside Merter: at the time of our research, the major production steps, such as sewing, usually happened in more peripheral parts of Istanbul, or, as in the first case (production for Western markets), also in other regions of Turkey or even abroad (Fig. 4.).
Fig. 4. Exemplary production chain of a knitwear production company in Merter, Istanbul, 2015.
Map: Anke Hagemann. Sources: Interview with firm owner, 2015, Google Maps.
27So while the ready-made clothes are distributed globally, most of the products are manufactured by cascades of subcontractors within the city region (Tokatli, Kızılgün and Cho 2011). Thus the district of Merter functions as a dense industry and wholesale hub with transnational significance. Initially, it developed as a district for small-scale manufacturing firms during the boom of Turkish export production in the 1980s. The upgrading of firms and relocation of manufacturing to other places began in the 1990s, while at the same time, the suitcase trade and wholesale functions – which expanded from the old town of Istanbul – started to take over specific parts of the district and have increased even more until today. Merter’s advantage has been, and still is, the high concentration of a diversity of textile firms, its proximity to related businesses in the metropolitan area and its good accessibility for international traders and buyers; it provides space for small enterprises and shops, but less so for large companies and mass production.
Fig. 5. Figure-ground plan of the Merter commercial area in its urban context, Istanbul, 2017.
Map: Anke Hagemann. Sources: Güngören Municipality, Google Maps.
28As an urban space, Merter is characterized by multi-storey commercial buildings arranged in dense blocks on a regular urban plan (Fig. 5). The district was originally planned for the relocation of timber merchants from the old town in the late 1960s, but when the individual buildings were constructed during the 1980s, they turned out to be very suitable and attractive for small textile businesses. The similar types of timber warehouses were turned into production buildings, often with one factory on each floor. These manufacturing premises were gradually adapted to other functions: headquarters, management, showrooms for clients or wholesale shops. This involved a broad range of physical transformations from individual extensions and modifications to complete facelifts and redesigns (Fig. 6-7).
Fig. 6. Typological study of buildings in Merter, Istanbul: Variations in the gradual transformation of facades and volumes, 2018.
Source: Anke Hagemann, Ilkim Er 2018.
Fig. 7. Typical street in the wholesale area for the so-called suitcase trade, Merter, Istanbul.
Source: Anke Hagemann 2015.
29Then, with the development of the suitcase trade, an entirely new typology evolved: Buildings were subdivided into small units for wholesale shops, often accessed through inner streets resembling shopping malls, while the facades are covered with shop windows and advertising. On a larger urban scale, the most striking changes are the upgrading and densification of Istanbul’s inner-city areas: Old industrial sites are successively redeveloped into commercial districts and large real estate projects are springing up around Merter. Hence, land prices are skyrocketing. This development is accompanied by planning efforts to push industry out of the city through land-use policies, among other regulative strategies (Hagemann 2015).
30Considering both network and urban space in synthesis illuminates the way in which they have affected each other mutually over time: The planning and construction of Merter initially provided spaces highly suitable to be adapted for the dynamic development of a textile cluster, and thus, the district became an important anchoring place in global clothing production networks. In the following decades, the economic dynamics and upgrading of the sector resulted in changes to the physical structure of buildings, while the limitations of these very structures – along with increasing land prices and development pressure in the city – contributed to the relocation of manufacturing to the outskirts, and thus to changes in the geography of production networks. In consequence, the upgrading of the industry and the built environment went hand in hand in Merter. At the time of our research, only very profitable businesses, mainly trade and producers’ headquarters, could afford to operate in Merter, while the district’s appearance was turning more and more into that of an upscale wholesale shopping zone. While this occurrence was largely unregulated, the municipality and a local business association have recently attempted to qualify the commercial development by giving selected streetscapes of Merter a corresponding facelift.
31While Merter started to lose its significance as a manufacturing cluster during the 1990s, the western hinterland of Istanbul was developing as a location for large-scale production, especially around the towns of Çorlu and Çerkezköy. At the time of our research, nearly half of the firms operating in the strongly industrialized region were based in the textile and clothing sector. These are usually branch plants of Istanbul companies and many of them are active in yarn or fabric production, including washing and dyeing processes. In particular, the biggest Turkish denim and jeans producers have factories around Çorlu. They are supplying major global brands, but most business operations – and often also the commodity flows – run through their Istanbul headquarters (Fig. 8).
Fig. 8. Exemplary production chain including a washing and dyeing factory near Çorlu, Western Turkey, 2015.
Map: Anke Hagemann. Sources: Interview with factory managers, 2015, Google Maps.
32However, in many cases, work-intensive production steps, mainly sewing, are performed by branches or subcontractors in Istanbul metropolitan region or in more distant places such as Kardzhali region, Bulgaria. As the factories in Çorlu are thus highly dependent on Istanbul and not really rooted or networked in their immediate local context, the place is appropriately described as a “satellite industrial platform” (Olcay and Erkut 2006) within global production networks. It was established as such in the 1980s and 90s, when space for production was becoming scarce and expensive in Istanbul. Çorlu offered attractive local assets including plenty of cheap land with excellent road connections, lower wages, good water resources for textile treatment and lax environmental protection policies. Moreover, by that time, Istanbul’s municipality sought to push out manufacturing from the inner city and Çorlu was identified and subsidized as an appropriate industry destination, even though it was and remains an important agrarian region. However, most of the dynamic relocation and expansion of industry to Çorlu happened without institutional initiation or planning guidance. Since the mid 2000s, several firms have again started to relocate production capacities from Çorlu to places with lower production costs, such as Anatolia, Bulgaria, Egypt, Jordan, or Ethiopia.
Fig. 9. Industrial and other built-up areas around Çorlu and Çerkezköy, Western Turkey, 2017.
Map: Anke Hagemann, Ilkim Er. Sources: Google Maps, OpenStreetMap.
33In terms of architecture and urban space, the production zones around Çorlu resemble suburban industrial areas in many other places of the world: The extended low-rise factory buildings are rationally and cheaply constructed, and of generic appearance. They sprawl along the main traffic routes on former farmland (Fig. 9-12). Rapid industrialization has also changed the face of the formerly rural towns, which have experienced intense growth. Most of the industry expansion happened without sufficient regulation, though, resulting in unplanned construction and conversion of agricultural land, heavy pollution and declining water resources. Only since 2010 have comprehensive spatial plans been implemented to restrict further pollution and industry sprawl, and to formalize the existing production areas as “Organized Industrial Zones” with the intention to provide the appropriate infrastructure.
Fig. 10. Industrial sprawl in Çorlu-Velimeşe, Western Turkey, 2017.
Map: Anke Hagemann, Ilkim Er. Sources: OSB Çorlu-Velimeşe, Google Maps, OpenStreetMap.
Fig. 11. Factory complexes in Çorlu-Velimeşe, Western Turkey, 2019.
Map: Anke Hagemann, Francesco Grillo, Ilkim Er. Source: OSB Çorlu-Velimeşe, Google Maps.
Fig. 12. Factory buildings in Çorlu-Velimeşe, Western Turkey.
Source: Anke Hagemann, 2015.
34Also in the case of Çorlu, a specific mutual transformation of translocal production networks and urban spaces can be identified. The material and spatial preconditions of the area played a crucial role as local assets in the initial coupling process. Subsequently, globalized mass production has had a massive impact on the physical landscape, to the effect of wasting some of the very resources and advantages of the location in the process of industrialization. The incipient desertion of Çorlu and the surrounding industrial region as a manufacturing platform for textile and clothing has been triggered not only by increasing production costs, but also by the more recent introduction of urban planning restrictions on harmful industrial sprawl and pollution. As clothing production is shifted elsewhere, some factory buildings around Çorlu are left underused or vacant, as aspirations to find new investors in “green” or more profitable industry sectors turn out to be hard to realize.
35One destination for the relocation of production capacities by firms in Çorlu is the south-east Bulgarian district around the city of Kardzhali, a predominantly rural region about four hours’ drive from Istanbul. Small and medium sewing factories involved in transnational clothing production are scattered in the small and middle towns of this district. In a network perspective, Kardzhali district figures as a modest production platform, where sewing subcontractors of transnational producers from Turkey, Greece and other European countries operate (Fig. 13).
Fig. 13. Exemplary production chains involving firms in Kardzhali, Bulgaria, and Çorlu region, Turkey, 2017.
Map: Anke Hagemann, Source: Interviews with factory managers, 2017.
36Industrial clothing production for export to the Comecon, also on contract for Western European firms, was established here in the state socialist 1970s and 80s. In the early to mid-1990s, small private firms set up sewing workshops and offered cheap outward processing opportunities for West European and Turkish clothing producers, drawing on the experience of workers discharged in the dismantling of the state socialist economic system. After a boom period of such outward processing, under often precarious conditions (Pickles et al. 2016, Smith et al. 2005), and since Bulgaria’s EU accession in 2007 (Evgeniev 2008), the clothing industry in the region has shown a mixed development up to the time of writing. A temporary consolidation of some firms is paralleled by the decline and shutdown of others, but also by on-going investments by several bigger Turkish companies to start new manufacturing operations in existing buildings. Closely-knit cross-border production networks have evolved, i.e. denim fabric produced and cut in Çorlu is sent to Kardzhali region for assembly, then the jeans return to Çorlu for washing and finishing before being exported to the shops and distribution centers of global brands in Western Europe. To some degree, companies form production sub-networks with flexible capacities in different towns across the region. But generally firms and government representatives assess the future of the clothing sector in the district as rather bleak, as the arsenal of cheaply available trained workers is shrinking, and further removal of production to Ukraine, Serbia, or African countries is under consideration.
Fig. 14. Industrial and other built-up areas in the city of Kardzhali, Bulgaria, 2018.
Map: Anke Hagemann, Ilkim Er, 2018. Sources: Google Maps, OpenStreetMap.
37In terms of spatial transformations, the initial establishment of clothing production in the 1970s meant the construction of some larger factory complexes in defined industrial areas according to modernist masterplanning, based on state socialist urbanization policy of infrastructure, industry and residential development (Fig. 14). The later outward processing boom did not trigger any major new industrial construction. Rather, disused buildings from the state socialist era – such as factories, army barracks, or department stores – were adapted in a basic manner, tapping into existing infrastructure networks, and only eventually did incremental conversion take place. Often, small companies would acquire and refurbish only sections or individual floors of buildings (Fig. 15-16). Depending on their economic success, and also their managers’ ability to secure EU grants for individual refurbishment, buildings and their equipment underwent gradual modernization, bringing them up to the current workplace standards of the global brands they supply.
Fig. 15. Partial conversion of vacant tool factory in Kardzhali to sewing factory and hotel, 2017.
Source: Anke Hagemann, Bucha Kelkar.
Fig. 16. Small sewing factory in a converted low-rise building (front center) in a residential district of Kardzhali, Bulgaria.
Source : Anke Hagemann, 2017.
38Thus in the case of Kardzhali, the significance of existing building stock as a local asset for “coupling”, to stay with GPN terminology, must be underscored – fixed capital from the socialist era, placebound but free floating alongside the human capital of skilled workers and production managers in the “transition” period. The availability of buildings suitable for sewing start-ups with low initial investment facilitated the (re-)insertion of this place into a web of transnational production arrangements, as producers in Turkey and elsewhere were seeking out options to lower production costs. This mutual transformation of production network and place amounted to a selective and unstable re-industrialization of a de-industrialized urban region. Up to the time of writing, the gradual adaptation of buildings enables the consolidation and geographic expansion of transnational production in this cross-border regional context.
39The emerging spaces of transnational clothing production in the fast-growing fringe towns of Addis Ababa stand in stark contrast to the small-scale companies established in the nooks and crannies of “post-socialism” in Kardzhali. Committed to attract export production in light industry with high labour intensity, the Ethiopian government has been seeking to push industrialization and infrastructure development, to generate jobs and much needed hard currency since the millenium, stepping up its efforts in cooperation with international, especially Chinese partners in recent years. Alongside preferential export agreements with the EU and US (since 2001), very appealing conditions for foreign direct investment were created, drawing scores of Turkish, East and South East Asian clothing producers in the years after 2005 to try their luck at setting up export production facilities in an environment with very limited previous experience of industrial development (Staritz, Plank, and Morris 2016, Beyer and Hagemann 2018). A number of large-scale factories and industry parks were established in the periphery of the fast growing country capital (Fig. 17-21) – among them direct relocations of production facilities from Çorlu, transformed into what was touted as the biggest textile and clothing factory complex of Sub-Saharan Africa for several years.
Fig. 17. Supply chains of global clothing factories near Addis Ababa, Ethiopia, as of 2017: a fully integrated knitwear factory in Turkish ownership in Sebeta (firm A) and a sewing factory in Bole Lemi Industry Park (firm B).
Map: Anke Hagemann. Source: Interviews with factory managers.
Fig. 18. Clothing factories in and around Addis Ababa, Ethiopia, 2018.
Map: Anke Hagemann, Elke Beyer, Rucha Kelkar. Sources: Google Maps, OpenStreetMap, africaopendata.org, ETIDI: Ethiopia Textile Sector Profile, 2017, IPDC: Industrial Parks Development Corporation, 2015, Addis Ababa City Administration.
40In terms of global production networks, these sites function as production enclaves of export processing zone type. Some plants are vertically integrated (from cotton preparation to finished clothes, see Fig. 19), but for the most part, major suppliers to global brands and retailers have set up sewing branch plants – essentially, extended workbenches to profit from the low costs for land, electricity, water and labour, incentives and tariff-free access to EU and US markets. Simple product types in large order quantities are the most common output, least sensitive in terms of lead time and required skills, but with extremely competitive pricing. To what extent industry clusters and supply networks between foreign firms within the country or even involving domestic firms can be formed, remains a decisive question of the future. The arrival of transnational clothing production in Ethiopia is a textbook case of “strategic coupling” hinged on the active interventions of an authoritarian federal government and involving not only transnational companies (buyers and suppliers), but also foreign and international development agencies as well as financial institutions as facilitators. But the rocky balance or foreclosure of several Turkish pioneer investors, at the time of writing, parallel to protest movements against what is perceived by some as government-assisted speculative “land grabs” falling short of the promised development benefits on a local level, indicates the high friction encountered in this endeavour to force an integration with global production networks. The lack or unreliability of infrastructural networks still in the making is among those challenges.
Fig. 19. Exemplary production sequence in an integrated knitwear factory near Addis Ababa, Ethiopia, 2017.
Drawing: Anke Hagemann, Ilkim Er. Source: Interview with factory manager 2017, Google Maps.
41The enclave character of the majority of sites of transnational production is very evident in an assessment from an urban and architectural perspective (Fig. 19-21). Large-scale monofunctional production facilities or manufacturing zones have been established by government-run industry park development agencies or private investors on land claimed by the government, with little apparent concern for an integration with the surrounding peri-urban sociospatial texture (Beyer and Hagemann 2018). They are privileged in terms of infrastructure provisioning for a cheap and steady power and water supply, as well as the speedy movement of commodities (Beyer et al. forthcoming), but urbanization effects – i.e. housing arrangements for the expected tens of thousands of workers and other aspects of social reproduction – appear only as an afterthought of planning, if at all. While the built environments of production comply with current workplace security and environmental standards of global buyer firms, they are nevertheless engineered to realize profits by very selectively connecting places across steep planetary economic divides, in the first place.
Fig. 20. Situation plan of Bole Lemi Industry Park, Addis Ababa, the first state-developed industry park for export production in Ethiopia, as of 2018.
Map: Anke Hagemann, Elke Beyer, Rucha Kelkar. Sources: Interview with park management 2017, Google Maps, IPDC: Industrial Parks Development Corporation, 2015, Addis Ababa City Administration.
42Thus, also in the periphery of Addis Ababa, a particular dynamic mutual transformation of urban space and network becomes evident. Large tracts of land or even fully developed building structures ready for plug-and-play were put at the disposal of transnational economic actors, combined with other local resources and incentives. Providing these spaces well-equipped for monofunctional use as transnational production enclaves is a primary aspect of the government’s strategic coupling efforts, resulting in the establishment of new anchoring points for global production networks. The current aspirations are to create a critical dynamic attracting ever more investment and more global production activities to inhabit the emerging industry parks and infrastructures, with promises for more comprehensive economic and urban development yet to be fulfilled. Meanwhile, the long-term and mediate urban effects of the new production enclaves are still unforeseeable.
Fig. 21. Bole Lemi Industry Park, Addis Ababa, Ethiopia.
Source: Elke Beyer, 2017.
43The four production locations discussed above form parts of overlapping global production networks, supplying the same global lead firms with similar products. However, the specific interaction between the globalized industry, physical assets and urban dynamics in each place has shaped urban production environments with highly diverse spatial characteristics: the dense inner-city textile hub Merter and the peri-urban industry sprawl around Çorlu stand in stark contrast to the dispersed manufacturing niches in Kardzhali region and the new large-scale enclaves for export production near Addis Ababa. Moreover, the intensity and scope of transnational influence vary: In all four production locations, the buyers, markets and standards of the produced goods operate internationally. But while in Istanbul region clothing production as well as the construction of manufacturing facilities are predominantly organized by domestic firms, the development of production locations in Kardzhali and Addis Ababa appears to be strongly influenced by foreign investors. Particularly in Addis Ababa, numerous transnational actors contribute to the production of the respective urban spaces – including not only global brands, but also clothing suppliers from South-East Asia, foreign development cooperation agencies and international financing institutions, as well as globally active infrastructure construction companies, banks and investors from the PR China. The varying outcomes of mutual transformation articulate the site-specific assets and trajectories, such as existing building stock or land to be developed, local labour availability and experiences in the textile industry, state incentives, and the specific activities and alliances of (trans)local governance and planning.
44A relational, multi-local perspective highlights not only interrelations, but interdependencies between places of production integrated within (or left behind by) global networks. From the 1980s to the time of writing, particular steps of clothing production have been shifted onwards from one urban and geographic setting to the other by firms in response and in relation to the particular conditions and local assets made available. The fine-grain analysis of these shifts has shown to what extent built structures (buildings suitable for manufacturing, and infrastructure networks for commodity circulation, power and water supply) contribute to creating relocation opportunities or pressures, and thus to reshaping the geography of production. Planning governance, i.e. changing rules and regimes for the use of space including environmental pollution, and governmental building activity to provide suitable infrastructure for the industry clearly emerge as crucial mediating factors to push out or pull in particular production activities. Our case study shows not only the volatility of globalized production, amounting to pronounced short term economic and physical transformations, but also its disposition to integrate and exploit extremely diverse localities simultaneously. Particular segments of production are competitively placed and displaced according to the specific character of particular places of production – based on the availability of spatial resources, i.e. land and/or fixed capital of building stock, infrastructures ensuring market connectivity, and other dynamic locational factors like labour availability and costs, firms’ qualification and production capacities, state incentivization and regulation. The relational nature of locational advantages, including the built environment and planning governance, emerges clearly in comparison of the different sites’ trajectories. Inner city districts of Istanbul such as Merter have developed from dense production clusters to supra-regional industry hubs and trade centers commanding parts of the production networks, while manufacturing has been outsourced to an expanding range of locations. Going hand in hand with global market dynamics, the spatial expansion and relocation of many individual firms has led to industrial sprawl in the metropolitan hinterlands around Çorlu, and partly spilled over to form ephemeral niches of production in de-industrialized Bulgaria – some of which depend on intense commodity circuits between Çorlu and Kardzhali. In only a short time, new “frontiers” of globalized clothing production have been explored by setting up export processing enclaves in countries like Ethiopia without much prior industrial experience. For instance, the creation of a huge industrial complex on greenfield land near Addis Ababa has left a textile factory in Çorlu vacant while company headquarters still remain in central Istanbul. Like a complex set of communicating vessels, economic and urban development in the four places appear inversely linked through relocation processes in the volatile industry with its intricate patterns of mobility and fixity. At the same time, integration with globalized production fuels multi-local processes of industrialization, de-industrialization and re-industrialization, de-valuation and re-valorization, or growth and decline. As a result, permanent insecurity and pressure to submit to the ever-increasing demands of transnational buyers affects every location of globalized clothing production, also in a physical sense. Most definitely, linear processes of “upgrading” coupled with comprehensive and inclusive economic and urban development must be seen as a rare exception from the general picture of ruinous competition and precariousness in diversity, amounting to the reproduction of disparities and uneven development on multiple scales.
45On the plane of theoretical considerations, research design and empirical evidence of a multi-local case study, we have highlighted connecting points and intersections between urban research and the GCC/GPN approach demonstrating the potential of these research fields to complement and cross-fertilize each other in multiple ways. GPN analysis yields instructive insights into the complex and multi-scalar constellations involved in the transnational production of urban space with most diverse site-specific outcomes. It also shows how different production locations are inversely linked through the volatility and mobility of the clothing industry. At the same time, our findings underscore the agency of the built environment and urban planning – whether as a crucial enabling factor or as a hindrance – in the dynamic formation and geographical shifts of global production networks. Applied to physical space, the notion of mutual transformation of place and network could thus be more precisely elaborated.
46The GPN approach can be usefully employed in urban research in order to take a relational and multi-local perspective on industry locations and the global economic processes they are part of. It allows to identify economic forces and actors on multiple scales which are involved in shaping the built environment, and to better understand the transnational constitution of the respective urban spaces. Further, it guides our view to rather inconspicuous spaces, and places that have been “off the map” as regards urban research so far. Thus, multi-local research along production networks provides ideal case studies for innovative, relational urban comparison (Robinson 2011) and for exploring the patterns of planetary urbanization (Brenner and Schmid 2015). Ultimately, looking at commodity circulation and firm relations represents another way to globalize urban research, one that has so far been neglected by global and transnational urban research approaches.
47Vice versa, an urban research perspective that pays close attention to architectures and material infrastructures, urban development and planning enables the grounding of global production networks in specific places and the built environment. By means of spatial analyses, the network nodes can be further unpacked and GPN research productively extended to the local level. Moreover, an urban research approach provides a way to investigate the potential significance of physical structures and urban planning for central GPN concepts such as embeddedness and strategic coupling, and thus, to explore the agency of material structures and urban development in the spatial (re)configuration of production networks. With an analytical view on urban territories, the understanding of regional development impacts can be expanded to aspects of urbanization and the transformation of physical landscapes, while at the same time this approach reveals some of the “dark sides” of global production networks, such as uneven spatial development, fragmentation, frictions, and conflicts on the ground.