Diversity in brokerage: how do gateway cities interlink their hinterlands?

ABSTRACT The article invites readers to rethink cities in economic networks against the backdrop of the many ways in which ‘gateway cities’, serving as brokers, interlink their respective hinterlands globally. It adds logistics, industrial processing and knowledge transmission to the more established gateway dimensions of corporate control and related service provision. This open heuristic is applied to Buenos Aires (Argentina), Cape Town (South Africa) and Singapore, which are vital nodes for the oil and gas industry. In addition to showing how diverse brokerage by cities is, the article calls for qualitative research to complement the insights on city networks generated by quantitative assessments.


INTRODUCTION
The world economy is characterized by the geographical fragmentation of production processes. Cities are crucial nodes that enable these complex processes by integrating dispersed locations into global value chains (GVCs). Studies that stand in the tradition of the Globalization and World Cities (GaWC) research network have made essential contributions to our understanding of globalization and cities in economic networks, in particular. In this literature, relations among cities of outstanding relevance tend to be emphasized (Derudder & Taylor, 2016;Taylor et al., 2002aTaylor et al., , 2002b. Cities that are topologically at the edge of the world city network, serving as brokers or 'gateways' and thus interlinking their respective hinterlands as commodity source regions or production platforms globally, meanwhile, have received considerably less attention. There remains a gap in the knowledge with regard to the functions fulfilled by these cities, as the diversity of core-to-periphery intermediation by urban hubs has not been investigated in a comprehensive and systematic way. In other words, we agree with an increasing number of colleagues who think that world cities should not only be analysed as part of the world city network but also as hinges between this network and the world beyond. We hence conceptualize cities as core-to-periphery brokersthat is, as intermediary nodes that interlink places that would not, otherwise, be well connected. By rethinking how such gateway cities integrate peripheral places into the world economy, we complement research on city-tocity relations in a globalized world, also reflecting on a recent trend in urban studies to place an emphasis on empirical diversity (Hoyler & Harrison, 2018).
The research on cities as intermediaries between global networks and the periphery of the world economy has generated important insights. Nonetheless, we agree with Martinus et al. (2015, p. 78) who observe that there is still a gap on 'how [not whether] cities … act as regional globalizing centers'. Related findings are mostly illustrative of core-to-periphery intermediation. They rest on induction and tend to remain a secondary topic in contributions that pursue other primary objectives. Advancing research that analyses the functional diversity of gateway cities in a deductive manner, we apply and further develop an open heuristic proposed by Scholvin et al. (2019). They suggest that gateway cities can be understood by taking five dimensions of global interlinking into consideration: logistics, industrial processing, corporate control and related service provision as well as knowledge transmission.
We stress that gateway cities are diverse, being marked by different combinations of the five dimensions, thus making a conceptual contribution to the state of the art, which does not sufficiently appreciate variety in city brokerage. In other words, we seek to uncover the diversity of a general phenomenon, answering the following guiding question: How do gateway cities serve as brokers, meaning as nodes in economic networks that interlink their respective hinterlands globally?
The remainder of the article is structured as follows. The next section elaborates on the heuristic of gateway cities and contextualizes it in debates on cities in economic networks. We show how the article fits into comparative urbanism and explain our case selection and methodology. The third section contains the analysis of Buenos Aires, Cape Town and Singaporethree gateway cities that play important yet different roles for the oil and gas industry. In the closing section, we point at topics for follow-up research that go beyond debates on city networks.

CONCEPTUAL FRAMEWORK, CASE SELECTION AND METHODOLOGY
Gateway cities A gateway city can be defined as 'an entrance into (and necessarily an exit out of) some area' (Burghardt, 1971, p. 269). Gateways serve as 'basing points' of capital that integrate peripheral places into the world economy (Friedmann, 1986, p. 69; see also Surborg, 2011). Because of this wide definition, research on gateway cities has become considerably broad, including topics very different from the one of this articlefor instance, urban dynamics that result from a city's intermediary role in economic flows (Grant, 2008;Grant & Nijman, 2002) or cities as channels and first destinations of international migration (Price & Benton-Short, 2008). Countless publications use the term 'gateway' but pursue research objectives other than clarifying or further developing the concept.
GaWC-inspired studies that conceptualize cities as gateways are focused on corporate headquarters in charge of business activities in hinterland locations (Alderson & Beckfield, 2012;Wall & Van der Knaap, 2012) and firms from the gateway city acting as global-to-local brokers (Martinus et al., 2021;Meyer et al., 2009). Martinus et al. (2015) as well as Sigler and Martinus (2017) distinguish between global networks and regional sub-networks of cities. Others assess how service providers integrate hinterlands into worldwide networks (Brown et al., 2002;Parnreiter, 2010;Zhang & Kloosterman, 2016). A distinction between 'decision cities' (for corporate control) and 'service cities' (for corporate services) has been made with regard to Brazil (Rossi et al., 2007) and applied to the oil and gas sector in Southeast Asia (Breul & Revilla Diez, 2017). Efforts to bring world cities and GVCs together also contribute to better understanding city-to-hinterland relations, as this literature is based on the conviction that world cities are vital nodes in GVCs, providing essential control and service functions for economic activities at peripheral sites (Brown et al., 2010).
We see two challenges to research along these lines. First, it is limited to corporate control and corporate servicesa rather narrow understanding of brokerage that has been criticized by post-colonial researchers because it amounts to 'reductionism [of cities] to only a small segment of economic activity' (Robinson, 2002, p. 545). Even though we do not identify with the post-colonial agenda, we agree that there is more to cities serving as brokers than corporate control and corporate services. We are aware of industry-specific studies that go beyond corporate services, thus revealing the diversity of the world city network (Hoyler & Watson, 2013;Krätke, 2014;Toly et al., 2012). This research refers to corporate headquarters and hence fails to address other forms of brokerage.
Second, we would not go as far as Watson and Beaverstock (2014) do. They argue that quantitative research on world cities has reached an impasse. However, we agree that qualitative approaches are able to shed light on features of cities in economic networks that remain beyond the reach of quantitative studies. Even Martinus et al. (2021, p. 13)who have shaped quantitative research on city networkspoint out that 'there is little empirical evidence of how [places] act as brokers or which specific connections they mediate'. Unlike these colleagues, we do not think that this gap of the knowledge results from a lack of data. We contend that qualitative research is necessary to uncover the complexity and exact ways of how firms use cities as gatewaysissues that have recently received more attention against the backdrop of calls to study agency and related processes of 'world city making' Parnreiter, 2014).
As the empirical analysis below demonstrates, qualitative research holds the potential to advance our understanding of cities in economic networks, being '"grounded" in the specificity of the individual processes … through which networks are formed' (Watson & Beaverstock, 2014, p. 419). It does so in a bottom-up manner (by starting at individual nodes in order to learn about the network), complementing the insights gained by GaWC-inspired quantitative research, which tends to operate from a top-down perspective (by beginning with the network and deriving conclusions on the nodes). This also allows us to grasp context-specific specialization of cities and the meaning of particular inter-city relations, which quantitative approaches have been able to map but, according to Watson and Beaverstock,unable to explain. 1 This article stands in the context of other contributions that go beyond the standard features from the world city literaturethose on 'interplaces' and 'relational cities' by Phelps (2017) and Sigler (2013), respectively. These studies provide significant insights. Sigler convincingly argues that spatial intermediation by cities is vital, but his article is less systematic regarding how intermediation plays out. Corresponding features of cities are revealed inductivelyfor illustrative purposes and without a claim of comprehensiveness. Phelps (2017, p. 10) Being convinced that gateway cities play a variety of roles as core-to-periphery brokers, we adopt a heuristic proposed by Scholvin et al. (2019). The heuristic originates from a review of the literature on cities in economic networks and concretizes Burghardt's aforementioned definition by suggesting five gateway dimensions: logistics, industrial processing, corporate control and related service provision as well as knowledge transmission. In other words, we have an ideal-type gateway in mindone that cannot be found in Phelps's and Sigler's contributionsand seek to learn about real-world diversity. The just mentioned dimensions are not necessarily additive. They can be combined in different ways (see also Scholvin, 2020aScholvin, , 2020b. We also consider the heuristic to be open. Explorative research such as ours may reveal further dimensions of gateway cities. Two aspects must be stressed for clarification. First, governance is a defining feature of world cities, but it is not necessarily one of gateways. A city that interlinks other places because it hosts a large harbour is not, usually, the site where corporate decisions on logistics are made. Second, gateway cities are end nodes of the world city network (Mans, 2013) (Figure 1). They connect to non-world city locations and whereas gateways feature advanced economic activities, the secondary cities and towns tied to them handle day-to-day operational tasks. In this regard, gateways differ in terms of their geographical range. For example, Mexico City appears to be a gateway to Mexico (Parnreiter, 2010), whereas Miami interlinks Latin America and the Caribbean globally (Brown et al., 2002).
It appears necessary to further explain how gateways are different from world cities, as there is an apparent overlap too. As Scholvin (2019) points out, the dimensions of logistics and industrial processing indicate that the gateway concept goes beyond the features that define world cities. Corporate control refers to regional headquarters that control subsidiaries at more peripheral sites (instead of global headquarters being in charge of worldwide networks). Service provision is about services needed in the hinterland. Knowledge transmission captures how global knowledge is adapted to local specificities and how local knowledge is made suitable for commercialization all over the world. It does not mean innovation in corporate services, which Sassen (2001) ascribes to world cities.

Comparative urban research
The empirical part of this article compares three cities, meaning that we follow Robinson's (2011) suggestion to analyse different forms of globalization in different geographical contexts. Comparative urban research has a long tradition. Many contributions seek to control for differences. They aim at concluding with a concept or theory that is generalized and presents cities in a homogenous way. Others appreciate variation. Adherents of the latter approach find it worthwhile 'to think cities through elsewhere', drawing 'on shared features … to stimulate … theorization across diverse outcomes' (Robinson, 2016, p. 191). With regard to our case studies, this means that the experiences of Buenos Aires, Cape Town and Singapore inform the conceptualization of gateway cities. They help to revise received assumptions.
At the same time, 'the process of theory building [in comparative urbanism is] more fragile and uncertain, and theory itself more unstable and less secure in its claims, as evidence from diverse and even divergent urban experiences will need to be engaged with and allowed to disturb conventional accounts' (Robinson, 2011, p. 17 (2016), we do not think that a cosmopolitan, grounded theory-like search for repetition and variation is the best way towards concept and theory building. On the contrary, we agree with Pierre (2005) who suggests to begin with a relatively robust modelthe one of gateway cities in our caseand adapt it in reflection of new empirical insights.

Case selection
Why have we chosen Buenos Aires, Cape Town and Singapore as gateways in oil and gas GVCs? Comparison makes sense whenever the objects that are to be compared have something in common but are not identical. We know that Buenos Aires, Cape Town and Singapore are world cities and thus critical to economic networks. 2 However, they are not alike. Singapore lacks oil and gas resources of its own. It is a highly developed city state, one of the 'regional command centres' identified by Taylor et al. (2002c) and appears to serve as a hub in oil and gas GVCs, as we gathered from preliminary desk studies. Cape Town is not a major world city. The wider hinterland -Sub-Saharan Africais largely underdeveloped, being deprived of many features that are essential for participation in GVCs such as reliable logistics and sophisticated service providers. Buenos Aires is the primary city of a hydrocarbon-rich country, whose economy is less liberalized than those of Singapore and South Africa. We focus on the oil and gas sector for two practical reasons. First, the sector is surprisingly transparent. Basic information on investments, ownership of/service provision to oil and gas fields as well as processing facilities is available online. We found lead firms and small enterprises to be accessible for interviews. They were willing to provide details on location strategies and cooperation with other companies. Second, oil and gas GVCs are marked by a rather simple linear configuration, as shown by Scholvin et al. (2019). The location of key assets and major players are easy to identify.
The sector also promises intriguing insights on gateway cities, particularly the concentration of oil and gasrelated activities there. The oil and gas sector is usually divided into down-, mid-and upstream. Upstream is about searching for resources, drilling wells and operating these wellsall of which is intensive in capital and technology. Not more than a dozen firms worldwide perform the most sophisticated tasks that oil field operators outsource. Each of them has to take decisions on where to carry out geological studiesthese are done remotelyand where to maintain and store equipment. These specialized suppliers subcontract other companies for less sophisticated tasks, having to choose where to seek these partners. Midstream comprises transport, storage and wholesale of crude or purified/refined products. Refining crude oil and purifying raw natural gas as well as marketing and distribution of products derived from oil and gas are the most important downstream activities. Again, lead firms have to choose suitable locations, either in proximity to oil and gas fields or elsewhere.

Methodology
Our analysis of Buenos Aires, Cape Town and Singapore began with desk studies to obtain general information and a first idea of key players and locational patterns that mark the oil and gas sector. The website A Barrel Full comprises information on all active oil and gas fields worldwide, indicating the respective operators (oil majors such as Chevron and Shell) and main service providers (Baker Hughes, Halliburton and the like). Information on the downstream sectorfor example, on refineries, their capacities and ownersis available too. Local newspaper articles were used to learn about recent developments. We also identified national and regional headquarters of companies mentioned on A Barrel Full via the respective company websites.
This sort of information provides valuable insights, but it is only a starting point. Details of how cities serve as brokers cannot be uncovered this way. In the course of a three-year research project, we therefore carried out 154 narrative, open-ended interviews with representatives of business lobbies, public authorities and, most importantly, local and transnational enterprises in the three cities and peripheral locations in Argentina, Indonesia and Vietnam (the latter two have resource peripheries that are globally linked by Singapore). The amount and variety of interviews allowed us to triangulate data and achieve a certain saturation, meaning that it is highly unlikely that we misinterpreted or overlooked important issues. Admittedly, the degree to which this can be demonstrated in a short empirical section is limited. There are also exceptions from the patterns we observed. We mention some of them further below.
The objective of the interviews was to better understand location strategies, inter-firm relations, intra-firm division of labour and location advantages that Buenos Aires, Cape Town and Singapore offer. A Barrel Full served as a starting point, but we identified most interviewees via LinkedInby searching on that network for people affiliated with important companies and key words such as 'oil', and specifying the location. Snowball sampling led to further contacts.
The interviews were conducted with the help of a guideline of eight questions that we adapted slightly before each interview, reflecting on the interviewee's area of expertise and the exact nature of his/her company or organization. We recorded the interviews and later applied standard means of qualitative content analysis (Mayring, 2015). We prepared selective and summarizing protocols, 3 structuring the interviews along previously defined categoriesthat is, the five dimensions of gateway cities. It is important to note that these dimensions were not introduced during the interviews. The corresponding insights therefore reflect the interviewees' understanding of Buenos Aires, Cape Town and Singapore in oil and gas GVCs, not ours. Qualitative content analysis allows for a revision of the categories that structure the empirical material, as a feedback from empirics to theory. This proved to be unnecessary for our interviews. We then chose anchor quotes, important examples and the related explanation by our interviewees for the empirical analysis that follows in the next section. Therein, we refer to 27 interviews (for details, see Table A1 in the supplemental data online).
The interviews not referred to contain fewer or identical information. No contradictory findings have been omitted.

THREE GATEWAY CITIES IN OIL AND GAS GVCs
In this section, we compare Buenos Aires, Cape Town and Singapore, referring to the five aforementioned dimensions of gateway cities: logistics, industrial processing, corporate control and related service provision as well as knowledge transmission. Each sub-section is structured accordingly. Table 1 summarizes our findings. The subsections further develop findings from earlier publications based on the same concept and methods (Scholvin, 2019(Scholvin, , 2020a(Scholvin, , 2020bScholvin et al., 2019). We have re-used some quotes from our interviews and insights based on data from secondary sources, and now focus on what is relevant to compare city brokerage.

Singapore: a multifunctional gateway
Singapore not only hosts one of the largest and most efficient ports in the world. The government of the city state also decided in 2006 to build Asia's first open-access, multi-user terminal for re-exporting liquefied natural gas (LNG). Because of that terminal, trading houses such as Trafigura and Vitol use Singapore as a distribution hub (Singapore LNG Corporation, 2018). What is more, equipment and service providers -Caterpillar and National Oilwell Varco, among othershave established spare part inventories there to supply sites all over Southeast Asia (interviews 1-3) (see also Scholvin et al., 2019), turning the city into a vital node in their own GVCs and those of their numerous clients. As Scholvin et al. (2019) summarize, resource-poor Singapore is, furthermore, the largest oil processing hub in Southeast Asia. ExxonMobil (592,000 barrels a day), Shell (500,000 barrels a day) and a joint venture of Chevron and PetroChina (290,000 barrels a day) have set up large refining capacities there (A Barrel Full, 2014a). A great variety of oil-based products are exported, mainly to Indonesia (23% of Singapore's corresponding exports), Malaysia (19%) and Australia (12%). Other countries in East, South and Southeast Asia account for 31% of these exports. A total of 53% of the refined petroleum imports of Indonesia and Malaysia come from the gateway city, 4 demonstrating its relevance as an intermediary in economic networks. Adding another example, Vietnamese blending plants source their lubricants from Singapore because such inputs are difficult to obtain on the domestic market (interview 4) (see also Scholvin et al., 2019).
The regional headquarters of several oil majors -Total, for instanceare in Singapore. Referring to the downstream segment, a representative of one of these firms explained that the office of his company in Singapore is responsible for corporate activities in 27 countries in the Asia-Pacific region and Middle East. It prepares the long-term strategy for these parts of the world, whereas the national subsidiaries handle the daily operational business such as customs issues (interview 5) (see also Scholvin et al., 2019). In other cases, Singapore matters even more for decision-making. An interviewee from Jakarta (Indonesia) told us that he and his colleagues only manage domestic sales on their own. They pass suggestions on investmentsthe construction of a blending plant, for exampleto the Asia-Pacific director, who sits in Singapore, and only if that person approves the plan, it will be forwarded to the global headquarters (interview 6) (see also Scholvin et al., 2019). Regional headquarters for upstream activities are also in Singapore, which serves as a 'coordinator and partly a conveyor' between the broader region, on the one side, and Europe and North America, on the other (interview 7) (see also Scholvin et al., 2019).
Singapore is of outstanding relevance for corporate services too. An interviewee from Jakarta stressed that: The city state moreover serves as a hub for maintenance services, in particular for offshore operations. A representative of an upstream company confirmed that many leading equipment providersthat is, European and North American firmssupply the region from the gateway city: 'if you need a wellhead, if you don't have a wellhead here in Indonesia, you will go to Singapore' (interview 9) (see also Scholvin et al., 2019). These firms also fly their specialized staff from Singapore to wherever they are needed in the Asia-Pacific region (interview 1) (see also Scholvin et al., 2019).
Further to that, Singapore channels knowledge. Most demonstratively, the company DNV GL established a deep-sea technology centre in the city state in 2012. DNV GL provides advice related to technical standards, their classification and development. The vice-president of the company said that the purpose of the centre was to 'spread [the] innovation genes instilled in [DNV GL] towards Asia' (Energy Boardroom, 2014).
Buenos Aires: a hub for corporate control Buenos Aires is the economic and political centre of a large country that possesses considerable conventional oil and gas reserves (2.4 billion barrels and 11.1 trillion cubic feet, respectively) as well as large unconventional resources. The latter are concentrated in the Vaca Muerta field in north Patagonia, which is one of the largest such deposits worldwide (Energy Information Administration, 2017).
Buenos Aires is Argentina's major container harbour and only hub for air transport. Most companies involved in the oil and gas sector are active in various parts of the country, as an interviewee pointed out. Travelling by air from one peripheral location to another usually involves a layover in Buenos Aires, making the city a preferred location for equipment and staff needed at more than one site (interview 10) (see also Scholvin, 2019). As a logistics hub, Buenos Aires connects to towns in close proximity of oil and gas fields, channelling global flows to places such as Añelo and Plaza Huincul in the province of Neuquén, which serve as secondary hubs, where frequently needed equipment is stored. Shipments of crude oil, conversely, pass through ports in Patagonia and Rosales, located at the edge of Bahía Blanca (Ministerio de Transporte, 2018). Scholvin (2019) observes that industrial processing concentrates in Buenos Airesbut only to a certain extent. Argentina's largest refinery, which reaches an output of 189,000 barrels a day, is in La Plata, a city about 60 km south-east of the centre of Buenos Aires. The second largest (110,000 barrels a day) is in Buenos Aires; the fourth largest (84,500 barrels a day) about 80 km north-west. Refineries in Bahía Blanca and in the provinces of Mendoza, Neuquén and Salta account for decentralization, indicating that Buenos Aires does not monopolize flows that interlink Argentina's downstream sector globally. Bahía Blanca also hosts a terminal for LNG imports. It is second in size to a similar facility 75 km north of Buenos Aires (A Barrel Full, 2014b, 2015. Plans to process the output from Vaca Muerta rely on Bahía Blanca as a hub, not on Buenos Aires (Río Negro, 2018aNegro, , 2018b. The gateway role of Buenos Aires is most pronounced in terms of corporate control. Argentina's capital is home to the headquarters of the semi-statal giant YPF as well as offices of all major oil field operators and service providers active in the country, including Pan American Energy, Weatherford and Wintershall. Key players from the downstream sector concentrate there too, turning Buenos Aires into the place where transnational firms make strategic decisions on activities in Argentina (within the limits imposed by the respective global headquarters). An interviewee explained that 'it's indispensable for a large corporation to be in Buenos Aires. … You know, we say, "God is everywhere, but his office is in Buenos Aires"' (interview 11) (see also Scholvin, 2019). Operational matters, meanwhile, are handled by offices closer to the resources, usually in provincial capitals (interviews 12 and 13). Large companies duplicate their organizational structures there (interview 14).
Numerous technical services are provided by local companies throughout the country. Legislation on local content, which refers to the provincial level, has proven critical in this regard. Beyond that, the director of an upstream service provider from the province of Río Negro pointed out that his company reacts much faster to any type of urgent necessity than external competitors. He added that being local and contracting local labour eases the resolution of conflicts, which are frequent in his home province, where oil and gas is extracted in the middle of fruit plantations and where unemployment is a significant problem (interview 15). Foreign service providers have invested in various locations across Argentina, either setting up subsidiaries there or buying local firms that complement their own capabilities (interview 16).
Companies from abroad also access knowledge in Argentina, for example, through institutes that carry out geological analyses or by employing highly qualified labour (interviews 17 and 18) (see also Scholvin, 2019). In hydrocarbon-rich provinces, some university programmes focus on the oil and gas industry. The town of Bariloche in Río Negro hosts research facilities that seek to plug into GVCs related to Vaca Muerta (interview 19). The Brazilian giant Petrobras, which sold most of its assets in Argentina in 2016, developed applied technologies in the city of Neuquén (interview 20). Yet, these are exceptions in a country where sophisticated research is heavily concentrated in the capital.
Cape Town: a specialized outlier Service providers such as Halliburton have rather large representations in Cape Town and fly their staff out to 220 Sören Scholvin et al. places all over Sub-Saharan Africa (interview 21) (see also Scholvin, 2020a). Cape Town's role as a logistics hub is related to the city being a services gateway. As Scholvin (2020a) explains, oil rigs used off the Atlantic coast are usually built in the Far East and, hence, pass by Cape Town on their way to Angola, Nigeria and other resource-abundant countries along the Atlantic coast of Sub-Saharan Africa. Many of these rigs are serviced in Cape Town because of the availability of skilled engineering companies, which are almost impossible to find elsewhere in the region (interviews 22 and 23). The Oil & Gas Directory, an online database set up by the South African Oil and Gas Alliance (SAOGA), indicates that Cape Town offers various services to the sector. They range from air charter to offshore engineering to travel agencies that handle visa issues, 5 and are used by clients from overseas, making Cape Town a place that enables them to integrate resource peripheries into global networks. A representative of SAOGA argued that 'we see a viable future here in the services industry, to serve as a hub for the region'. He continued to explain that the development gap vis-à-vis the subcontinent means that there is a lack of sophisticated services beyond South Africa's borders, stressing that '[this] is a major selling point [for Cape Town] that we are trying to make' (interview 24) (see also Scholvin, 2020a). There are, of course, exceptions from this pattern. A former employee of the service provider Schlumberger explained that since the end of Angola's civil war, the company has relocated much of its equipment from Cape Town to Luanda (Angola) (interview 25) (see also Scholvin, 2020b). Scholvin (2020a) notes that the concentration of service providers is boosted by the development of Saldanha Bay, located 120 km north of Cape Town. Saldanha Bay is a natural deep-sea port. In 2013, the Saldanha Bay Industrial Development Zone (SBIDZ) was launched as a cluster for oil and gas services as well as marine repair. 6 Still, the SBIDZ is in its initial stages. It constitutes a prospect for Cape Town, rather than a present-day reality.
With some restrictions, Cape Town is also a hub for industrial processing, but it does not play a corresponding gateway role yet. A refinery in the suburb of Milnerton, formerly owned by Chevron and bought by Sinopec a few years ago, reaches an output of 110,000 barrels a day (A Barrel Full, 2014c). Cape Town used to be one of Chevron's three global supply hubs. Before the Covid-19 pandemic, the new owners made somewhat vague announcements on major investments, which might enable Cape Town to export more refined products to the neighbouring countries (Engineering News, 2018;Times Live, 2018).
Major oil companies -Anadarko, Eni, Tullow and othershave offices in the city, but they are not regional headquarters as in the case of Singapore. According to an interviewee from one of these firms, Cape Town 'has a door function' for South Africa, meaning that it serves as a corporate gateway on the national scale, at best. Subsidiaries of the same company in other Sub-Saharan African countries interact directly with offices in Houston (USA) (strategic decisions) and London (UK) (technical support) (interview 26) (see also Scholvin, 2020a). For Tullow, this is somewhat different because Tullow bought an exploration company called Energy Africa in 2004 in order to expand into the subcontinent. Energy Africa happened to be based in Cape Town. Most of its staff stayed there, working on exploration studies for the new owners (interview 25). A director of a spin-off explained that his firm takes small equity in technically complicated exploration projects all over Sub-Saharan Africa, being a partner of investors from overseas, whose input is limited to finance (interview 27). As Scholvin (2020a) concludes from these observations, it would be difficult for the investors to succeed without the region-specific expertise available in Cape Town. Hence, the city plays a considerable role for knowledge-intensive components of some oil and gas GVCs.

A LOOK AHEAD
This article advanced research on city networks conceptually and methodologically. The main objective was to show how gateway cities interlink their respective hinterlands globally. Our analysis revealed the diversity of cities that serve as brokers and also demonstrated the usefulness of qualitative research. It encourages future studies to broaden the conceptualization of cities in economic networks and complement insights generated by established quantitative methodologies with findings that, in our opinion, only result from qualitative approaches.
A key merit of the comparison of three cases is that it revealed that gateway cities are not necessarily like Singapore, which might, otherwise, be misunderstood as some sort of role model, especially because Scholvin et al. (2019) refer only to this city to exemplify their heuristic. Singapore is a multifunctional gateway. It serves as a logistics hub and is a major site of industrial processing. The city state hosts regional headquarters of lead firms. Corporate services and important research facilities concentrate there. Buenos Aires can best be described as a hub of corporate control. It bundles logistics and knowledge transmission to a certain extent, whereas industrial processing and technical services are more decentralized across its hinterland. Cape Town is a specialized outlier, vital for oil and gas GVCs because of logistics and technical services. The city has a certain relevance for knowledge-intensive inputs for the upstream sector and might become a refining hub. Corporate control, meanwhile, does not characterize Cape Town's gateway role.
A feature that the cities analysed in this article have in common and that should be taken up in follow-up research is that they remain in a subordinate position on the global scale. Singapore hosts major transnational companiesbut only their regional headquarters. Transnational firms in Buenos Aires organize oil and gas GVCs across Argentina. They are bound to decisions taken in Europe and North America (interview 28). Cape Town's prosperity as a logistics and services hub depends on whether Anadarko, Eni and others contract Cape Diversity in brokerage: how do gateway cities interlink their hinterlands? 221 Town-based enterprises. They may opt for alternative ports such as Las Palmas on the Canary Islands (Spain) or Walvis Bay in Namibia. Tullow could centralize its knowledge-intensive activities in Dublin (Ireland) or London, closing the South African office (interview 25). The fact that Buenos Aires, Cape Town and Singapore compete with other cities points at another topic that merits more attention. Cities in a globalized and liberalized world economy must convey why they are relevant and valuable business options. They compete for investment and skilled people (Kavaratzis, 2004;Kavaratzis & Ashworth, 2005). This article focused on location strategies of companies that turn some cities into gateways. They are, at least to a certain extent, influenced by public policies. How these policies are drafted and whether they are successful in creating and reinforcing gateways should be investigated. Critical assessments could tie up with the critique of creative class and world city-agendas (Peck, 2005;Robinson, 2002).
Further to that, follow-up studies are needed to explain why Buenos Aires, Cape Town and Singapore assume the specific roles that we observed. We suppose that Singapore bundles many gateway functions because it is one of the most advanced economies, located among much less developed countries. The government pursues a proactive strategy of positioning the city state as a hub in GVCs (Breul, 2019). The development gap between South Africa and the rest of Sub-Saharan Africa is also considerable. Unlike Singapore, Cape Town cannot outcompete overseas hubs of corporate control though. Buenos Aires is a primary city that interlinks a national hinterland. The local density of key playersprivate and publicis essential (see also Scholvin, 2019). Some segments of oil and gas GVCs are carried out in other parts of Argentina, probably because these territories have a long track record in hydrocarbons.
Finally, many GVC-related activities beyond resource extraction can be carried out at peripheral locations, as the Argentinean case study showed, but this is not always the case. Breul et al. (2019) argue that Indonesia and Vietnam suffer from Singapore's 'filtering mechanisms'. The gateway city absorbs tasks that could also be done in resource peripheries. There is, hence, need for a more thorough analysis of the dark sides of interlinking by gateway cities oreven betteran assessment of what resource peripheries can realistically expect from integration into global networks.

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.

NOTES
1. Qualitative research apparently also suffers from shortcomings. Conclusions drawn from two or three cases cannot, usually, be generalized. The comprehensive datasets used in quantitative studies are often seen as more reliable than statements by a few interviewees. Mixed-methods approaches are a sound way to combine the strengths of qualitative and quantitative research. They proved feasible for one of the regions we are interested in (Breul & Revilla Diez, 2017). 2. For the latest GaWC rating, see www.lboro.ac.uk/ gawc/world2018t.html. It concludes that Buenos Aires is an 'alpha' world city; Cape Town reaches 'beta' status; and Singapore is rated 'alpha plus'. 3. In order to prepare selective and summarizing protocols, the researcher decides to skip irrelevant parts of the interviews, while listening to the respective audio files. The method is, hence, much more focused on pregiven topics of interest than full transcription. 4. Calculations are based on data obtained at https:// comtrade.un.org. 5. SAOGA is a business lobby. For the complete list of companies, see www.saoga.org.za/directory. 6. For this and further information, see www. saldanhaindustrial.co.za and www.sbidz.co.za.