GaWC Research Bulletin 175

This Research Bulletin has been published in Geoforum, 38 (4), (2007), 628-642.

doi:10.1016/j.geoforum.2006.11.005

Please refer to the published version when quoting the paper.

(Z)

Transaction Links through Cities: 'Decision Cities' and 'Service Cities' in Outsourcing by Leading Brazilian Firms

E.C. Rossi, J.V. Beaverstock and P. J. Taylor*

Abstract

Most world city studies focus on where the provision of advanced producer services takes place, in this study the location of the service clients are also taken into account. Production-service transactions are identified as links between ‘service city’ and ‘decision city’ (client HQ) and classified as intra-city (where a firms uses local (same city) services) and inter-city which may be domestic or foreign depending on whether the client uses services in or out of the country. Sampling the leading firms in Brazil, 218 transaction links are described and analysed in terms of cities, firm sectors, ownership and market. The analyses are used to inform a rereading of gateway city functions that takes into account the concentration/dispersal nexus in economic globalization. Thus whereas the dominance of Sao Paulo in the Brazilian economy is confirmed, this is not interpreted as a simple gateway function with this one city ‘articulating’ the national economy into the world economy. A much more complex situation is revealed with by-passing the gateway also much in evidence. The conclusion is that this shows concentration and dispersion of economic activity happening simultaneously in Brazil.

Keywords

Advanced producer services, Brazilian cities, gateway city, globalization, production-service transactions, world cities

INTRODUCTION

This is an empirical paper that builds upon the urban geography and economic globalization literatures at three levels. First, we add to the slowly growing number of studies of ‘non-core’ cities in the world city literature. Still relatively neglected, we continue our contribution to alleviating this important geographical limitation on research by focussing on Brazilian cities in globalization (Rossi and Taylor 2005; 2006). Second, we explore the gateway city concept as a means of making sense of how countries like Brazil are connected into the wider world economy. Again following our previous studies (Rossi and Taylor 2005, 2006), we know that ‘gateway functions’ are much more complex than the gateway city concept implies; here we illustrate and explicate these processes in new ways. Third, these new ways are through the study of transaction links, bringing together the geography of the users of professional services with the geography of their service provision. This empirical linking of service providers and their clients is a new research departure for world city studies and therefore constitutes the originality of this paper.

In an important critique of the world cities literature, Robinson (2002, 536) complains that ‘millions of people and hundreds of cities are dropped off the map of much research in urban studies’. This exclusion is from two ‘maps’: the geographical map of world cities wherein most cities in the ‘South’ are missing; and the conceptual map of world cities which focuses on a narrow range of global economic processes so that myriad other connections between cities are missing. Basically she complains that a few ‘Northern’ cities have been used to set a world city-ness standard that has then been applied to ‘Southern’ cities which have consequently been found wanting. She calls for a post-colonial approach to create a ‘more cosmopolitan urban theory’. Subsequently (Robinson 2005, 760), she has conceded that this literature now covers ‘a much wider range cities around the globe’ thus lessening the exclusion from the first map. In this paper, we extend the range even further by studying some Brazilian cities that have never previously appeared in the world city literature. Thus we make a real contribution to putting a few cities at least ‘back on the map’.

As initially mentioned, this is an empirical paper and therefore we do not directly address Robinson’s conceptual map exclusions. Following Sassen (1991/ 2001) we continue our focus on advanced producer services but we do place these in a new conceptual context that does recognise Robinson’s emphasis on the inherent complexity of cities (Robinson 2005, 757). In our detailing inter-city links within and without Brazil, we contribute to unravelling some of the intricacies of flows that hold the world economy together, including Brazil’s place within it. Thus we go further than the many case studies of non-core cities in globalization (e.g. in edited volumes such as Lo and Yeung (1998), Aguilar and Escamilla (1999) and Gugler (2004)) wherein the researches reported do not directly show how the subject cities actually link into a globalizing world economy. Very often the focus on one city leads to concentration on local adaptations to globalization processes – an important topic – but without concern for the whats and wherefores of the causes of the disruptions that require adaptive responses. By focussing on cities across a large country – Brazil – we are able to provide two important additions to studying cities in globalization: first, a national contextualisation for the operation of globalization processes, and second, a critical evaluation of the use of the gateway city concept that tends to privilege concentration processes over dispersion in understanding contemporary globalization. Each of these additions have been developed in detail in our previous papers on Brazilian cities in globalization; to avoid repetition we do not go over exactly the same ground again, rather we summarize the key points in order to make this paper self-standing.

Five key points need to be kept in mind when considering Brazilian cities in contemporary globalization (for fuller exposition, see Rossi and Taylor (2005) and Fernandes and Valença (2001)).

1. Powerful integration of the Brazilian economy into the world economy through city networks long predates contemporary globalization. IPEA et al (2001, v.2, p. 346) identify three previous urban-economic stages: Atlantic slave territorial formation, agricultural-market territorial formation, and national-urban territorial formation.
2. Through the 1950s to the 1970s the latter formation created an industrial process and endogenous capital accumulation with an associated national urban network (IPEA et al., 2001, v2, 346). Thus Brazilian cities were firmly embedded in a city-based industrial economy before contemporary globalization.
3. Interpretation of the national-urban territorial formation and how it relates the contemporary era is contested.
4. For Milton Santos (1993), the state lost control of national territorial planning with the economic collapse of the 1980s. Major firms, both domestic and foreign, took charge of the national market creating a highly vertical structure centred on Sao Paulo. Development of this new economic geography has continued into the 1990s (Fernandes and Negreiros 2001, 425, 428).
5. IPEA (et al.2001, v.2, 346) offer an alternative understanding of this process that they term an international financial phase starting in the 1970s through to the 1990s. Here the exhaustion of the financial model of the state has resulted in dispersion rather than concentration, for instance in new dynamic centres such as Manaus, Brasilia-Goiania and Fortaleza.

This dispute between economic dispersal and concentration in Brazil is mirrored in other countries and is central to our exposition of gateway functions in contemporary globalization.

There are six key points that underpin our use of the gateway city concept (for further details see Rossi and Taylor (2006)).

1. In the traditional use of this concept ( Johnston 1982), gateway cities were contrasted with central places (Ullman 1957); they violated the latter’s spatial symmetry. Such a city was seen as a combined entry/exit for a specific region (Burghardt 1971), historically linking a frontier region to distant cities (Drennan 1992).
2. A second strand of work emphasized gateway cities as communication hubs (Ullman 1957), latterly crossroads for information flows (Pred 1977). Both these traditional uses – links to distant cities, information flows – suggest relevance to contemporary globalization (Short et al 2000)).
3. But in translating these ideas to conditions of contemporary globalization there has been a tendency to tell only half the story; the ‘region’ being exited/entered is neglected. For example, in Friedman’s (1995) identification of world cities as ‘national articulations’ one side of the articulation – world cities in the world-economy – is described in some detail but the ‘national economies’ being ‘articulated’ are treated as given. (See also Andersson and Andersson (2000))
4. This limitation is found in three very significant papers on non-core cities in globalization: Grant and Nijman (2002) observed that the gateway function of world cities in less-developed countries is operated through the large port cities (Accra and Mumbai), but there is little mention of these cities’ hinterlands; and in a related paper, Grant (2001) provides detailed descriptions of foreign markets linkages through the foreign companies but, again, Accra’s hinterland, by implication the rest of Ghana, is not similarly treated. Similarly, Olds and Yeung’s (2004) ‘emerging global city’ construct has a one-way insertion: from the global arena to the emerging global city, connoting its ‘entry’ function as global city; this emerging global city has no hinterlands.
5. Thus the use of the gateway city concept in globalization studies commonly assumes concentration in economic processes but, because the focus is on just half the gateway flows, important questions remain open: how dominant are these gateway cities within their own ‘national economies’? or, put another way, how dependent are other cities on their country’s ‘gateway city’?
6. However there is one use of the gateway concept that allows for economic dispersion: for Short et al (2000) studying gateway cities within countries can be a way of extending the range of cities researched in globalization. They ‘use the term gateway to refer to the fact that almost any city can act as a gateway for the transmission of ... globalization’ (p. 318). For instance, the small US city of Sioux Falls is one of seven ‘sample cities’ they discuss.

In our study of Brazilian cities in globalization we study cities inside and outside Brazil and in the former, as previously noted, we treat cities new to globalization studies.

This contrast between economic processes that both concentrate and disperse activities is not seen as a contradiction in the world cities literature. Rather it is, according to Sassen (1994, 20), a ‘central proposition of the research literature on global cities’. From the beginning with Friedmann’s (1986) linking the new international division of labour to the need for ‘command and control centres’ through Sassen’s (1994, 20) explicit recognition that ‘territorial dispersal of economic activity … has created new forms of concentration’, globalization processes have been interpreted as simultaneous dispersion and concentration in the world cities literature. What does this denote for the gateway concept in contemporary globalization? First, the concentration process means that there will be gateway city formation. Second, the dispersal processes mean that there will also be by-passing of the gateway; other cities can prosper directly as a result of globalization. Thus, contested interpretations of the current economic geography of Brazil are potentially explicable: powerful concentration and dispersal processes occurring simultaneously imply that the difference of opinions given above is not so problematic after all. The purpose of this empirical study is to explicate these processes for Brazil and thereby show how the gateway city concept can be conceptually embedded into world cities studies, especially for non-core regions.

Transactional links in servicing production

Following the seminal study of global cities by Sassen (1991/2001), inter-city relations in globalization have been dominated by study of advanced producer services. For instance, the world city network has been defined as multiple interactions treating cities as global service centres (Taylor 2004). Our previous two studies of Brazilian cities in globalization have continued in this tradition (Rossi and Taylor 2005, 2006). It is this focus on studies of advanced service provision that is criticized by Robinson (2002) as too narrow for adequate understanding of contemporary cities. However, we do not treat such advanced service firms as dominant players – service corporations are by no means the largest TNCs in the globalizing world economy – rather we see them as critical indicators of the operation of economic globalization processes.¹ Nevertheless in this study we do extend our empirical gaze beyond service providers: simultaneously we look at the corporate users, their clients, as well.

This way of thinking requires merging of two distinctive spatial models of economic globalization: the world city network and global commodity chains (Brown et al 2006). The latter constitute a sequence of nodes of production where labour and means of production are brought together to add value to commodities en route to final consumption and capital realization (Gereffi and Korzeniewicz, 1994; Pelupessy, 2001). All this production requires various and different professional services at each node, inter-relations between the agents involved from procuring inputs, through production, to placing outputs on the market (Dicken 2003). This involves a wide range of activities that mix production with services that can be organised in many ways. The role of advanced producer services, however, is relatively straightforward. As highly specialised financial and professional services, they have two properties critical for our research: first, they are largely outsourced by the corporations organising production chains; and second, the services are largely produced in world cities. Thus the world city network envelopes commodity chains through providing strategic inputs across its whole length from capital for start ups, to the final advertising for selling the product with, for instance, advanced legal, insurance, accounting and management consultancy inputs along the way. It is these service-production transactions that are the empirical subject matter of this paper.

Ideally we would have liked to study complete production chains but as an initial empirical foray into this field we have treated single individual service-production transactions as our units of analysis. The specific transaction links that we study are between the corporation headquarters where decisions on outsourcing are made, and the particular offices of service providers where the financial/professional/creative work is carried out. Such transaction links operate through cities: for every link there is a ‘decision city’ where corporate headquarters commission the services, and a ‘service city’ where the servicing occurs. This can produce both intra-city transactions, where decision and service are in the same city, and inter-city transactions, where decision and service are in different cities (Figure 1). It is not always appreciated that in Sassen’s (1991) original formulation of her global city concept, these major cities were interpreted as markets for services as well as locales of service provision. This is represented in Figure 1 as intra-city transactions where decisions are made to use local (same city) services. In fact, of course, not all headquarters are in major cities (Lyons and Salmon 1995) and therefore outsourcing may necessitate using services in another city that is larger and provides more advanced services. Also, very often TNCs make international outsourcing decisions, perhaps to known and reliable service firms in their own country. Figure 1 illustrates both of these inter-city transaction types.

Since Sassen’s (1991) emphasis on the advanced producer services in global cities, interest in Friedmann’s (1986) original focus on world cities as ‘command and control centres’ has greatly diminished. This is largely because pressures for locating in world cities are much less compelling for headquarter functions than for professional service provisioning. Thus the vast majority of world city research deals with our ‘service cities’ but not our ‘decision cities’. Given this situation, why should we resurrect a largely abandoned interest? Why are decision cities of interest for understanding economic globalization? The general answer is, as Dicken (2003) has long emphasized, that the nature of firms varies greatly in terms of their origins and this is of prime importance to non-core countries where foreign firms may be dominant in globalization. Within countries, especially large countries like Brazil, this may be extrapolated down to the inter-city scale (intra-national differences) but this is not a major reason for bringing decision cities back into focus here. There are two important reasons to study decision cities. First, there is the obvious policy issue that where decisions are made about deploying resources and making creative work cannot be ignored. The practical result is that cities can be identified as either ‘service-dependent’ or ‘service-sufficient’. A service-dependent city is a decision city where firms tend to outsource service to other cities. A service-sufficient city is a decision city where most outsourcing stays in the city. Such cities may also import outsourcing from other cities. This latter process leads to the second reason, exploring the salience of the gateway city concept in globalization. If countries are linked to the world economy through ‘gateways’, this should be especially prevalent in advanced producer services as key indicators of economic concentration in globalization. It is at this conceptual level, that decision cities are necessary for this study: this is where the economic agents are who are choosing to concentrate or disperse services in a globalizing world economy. To what degree are ‘service-sufficiency’ outcomes for cities in countries of the ‘South’ concentrated into just one decision city, the gateway city?

Leading firms in Brazil and the servicing of their production chains through cities are the subjects of our research. The data consists of 218 transaction links between 8 decision cities in Brazil, and 26 service cities, both inside and outside Brazil. We are interesting in describing the geographical configuration of these transaction links to assess the operation of the gateway concept in Brazil. This is accomplished in three substantive parts. The first part outlines the data collection of the transaction links and characteristics of the firms in the sample. The results are then presented in two parts. The second section deals with the aggregated data for all firms; decision cities and service cities are identified and their links described. The third section disaggregates by firm characteristics: firm sectors, service sectors, firm and service ownership, and whether the servicing is for a firm’s domestic or foreign business. Finally, in a brief conclusion our results are interpreted in terms of their meaning for production chains and world cities beyond the Brazilian case study.

Data collection

Studies of advanced producer firms in world cities have generally focused upon the location decisions of the providers of the services – which cities they choose to locate their offices. In this study we investigate the decisions of the users of the services – where they choose to be serviced. Such information is far more difficult to obtain; it requires a detailed activity survey of individual user firms. We collected data on transaction links, in Brazil or worldwide, that major firms operating in Brazil create through their use of advanced producer services. The basic method used was an e-mail survey followed up by telephone calls. The latter turned out to be vital; phone calls became the main source of data.

Emails were sent to Brazil’s 120 top firms as listed in the printed edition of EXAME Melhores e Maiores of 2003 (data refer to 2002) between 5 th April and 3 rd September 2004.² The e-mail messages were written in English and included 24 questions. Email addresses of the chairman of every firm were obtained by telephone, and each was invited to take part in the project. A total of 18 firms³ agreed to co-operate. The cities in which their headquarters are located were identified as ‘decision cities’, i.e., the places where decisions on outsourcing the servicing are made (‘TNC’ on Figure 1).

Information obtained from the questionnaire was basically on the geography of service use. Each firm was asked to identify the main firm used for the following services: (i) accountancy, (ii) advertising, (iii) banking (creditor bank), (iv) law (commercial or economic legal services), (v) insurance, and (vi) management consultancy. The specific office used was obtained so that where the servicing was carried out was found – thus we identified ‘service cities’ (‘APS’ on Figure 1). This information was collected for domestic and international business separately. Not all user firms provided information for all services, and some identified more than one service firm for the same service. The final result was 218 specific examples of outsourcing services, transaction links identifying a decision city and a service city.

In addition, further information was sought on characteristics of the agents making the transaction. For the service users, three key pieces of information were gathered: (i) the business sector in which the firm operated, (ii) ownership of the firm (shareholding control of the company), and (iii) the scale of the transaction (domestic/international). For the provider firms, the service sector and the world headquarters were recorded. The total data set provides a unique source for exploring transaction links through cities.

Tables 1 and 2 summarise the data and relates them to the initial population of 120 firms. Table 1 constitutes a check on the representative of our sample of firms in terms of the two basic characteristics of user firms, their city location and their business sector. First, Table 1 shows that our sample is much more geographically concentrated than that of the initial population of firms. The latter’s headquarters are spread across 24 city regions but in our sample there are only 8 cities feature. This is obviously an effect of our relatively small sample of firms, which we were unable to increase. However given just 18 firms, finding 8 decision cities is a reasonable outcome, especially because they are the top 8 cities in Brazil in terms of economic importance (see appendix 1 in Rossi and Taylor 2006).

Although the firm sample size is relatively small, the basic unit of information we analyse is the transaction link of which there are 218 and these have been included in Table 2 . The first three rows show the distribution of firms by aggregated sectors. The initial 20 business sectors reported in Table 1 are reduced to three essential sector categories: (i) Traditional (automotive, building, gross and foreign trade, mining, paper and cellulose, retail sales, steel and metallurgy); (ii) Innovative (chemistry and petrol-chemistry, electrical and electronic appliances); and (iii) Basic (public services).⁵ The final four columns divide firms by ownership in terms of private, state and foreign control. The first important point that Table 2 makes is that our sample of 18 firms is represented across all categories of sector and ownership. Most importantly, the final column shows that we have reasonable numbers of transaction links across all sector and ownership categories – the lowest numbers are 17, 27 and 47. Thus we conclude that the total of 218 links is enough to credibly support an exploratory analysis of Brazilian transaction links through cities.

The geographies of transaction links

The overall results of this research are shown in Table 3 . Besides the 8 decision cities, 26 service cities are recorded including the majority, 16, from outside Brazil. The following three general findings are derived from this table.

• Of the 218 transactions, 113 decisions (52%) are made in Sao Paulo and Rio de Janeiro, and 132 service provisions (61%) are in these two cities. That these two cities dominate Brazilian transaction links is not surprising; showing them to be leading more as service centres than as headquarter cities does initially suggest gateway function processes.

• Of the 218 transactions, 135 are intra-city links (62%) and 83 are inter-city links. We are unaware of any other research that compares these two types of transaction links and so we have no context for interpreting this result.
• Of the 83 inter-city links, 33 (40%) are with foreign cities. However, consider that nearly a third of these (10) are transactions using services in New York. Further, the dominance of New York in foreign servicing is strongly confirmed by the fact that decision-makers in five Brazilian cities have chosen this service city, whereas no other non-Brazilian city is chosen in more than one Brazilian decision city.

To search out more specific findings, in all remaining analyses we focus on the 8 decision cities and aggregate choices into the three categories illustrated in Figure 1: intra-city transactions, and the two forms of inter-city transactions, domestic and foreign. This simplification is made for two reasons: first, to cut through the detail shown in Table 3 and second, to take account of our relatively small sample size – 10 or more choices are recorded for only 5 of the 26 service cities (Table 3).

Table 4 shows the distributions of transactions across the three choice categories for the eight decision cities. Since numbers of choices made in each of these cities varies greatly, the distribution is also shown as percentages. Two straightforward findings illustrate the salience and meaning of these results. First, as Brazil’s leading city, it is perhaps not surprising that Sao Paulo records most self-selections, and second, as state capital (and very much a ‘specialist political city’) Brasilia records the least choices outside Brazil, zero in fact. The only other zero recorded in Table 4 is for Porto Alegre having no transaction links with other Brazilian cities. This emphasizes the city’s national separateness, a feature shared with Salvador, which has just one transaction link to other Brazilian cities. Also, these two cities rank just below Sao Paulo and Rio de Janeiro for intra-city transactions. The transaction separation of these ‘second-tier cities’ is most likely due to their respective domination of important economic regions away from Brazil’s economic core-region.

Adding to these initial observations, a cursory analysis of Table 4 shows that each characteristic divides the cities into two groups: the intra-city percentages have a gap between 36% and 62% to identify ‘self-sufficient service cities’ and ‘service dependent cities’; for ‘other domestic choices’ the gap is between 22% and 46%; for the ‘foreign choices’ the percentages gap is between 12% and 22% in low and high categories. Using the intra-city choice and foreign divisions a simple typology of decision cities is created (Table 5 ). Thus Sao Paulo and Rio de Janeiro form a city-type with high intra-city choice and low foreign choice reflecting their national economic prowess as the leading cities for both decisions and servicing (Table 3). The largest city-type combines Brasilia, Curitiba and Campinas, which are low in both choice categories indicating service dependence on other Brazilian cities. Curitiba and Campinas actually use Sao Paulo services more than their own city services (Table 3). Brasilia is distinctive in spreading its service provision: although we have only 9 transaction links for this city these cover 6 Brazilian cities, two of which are non decision cities. Belo Horizonte forms a singleton type combining low intra-city choice with high foreign choice. Turning to Table 4 we can see this distinctiveness relates to the city’s distinctive, relatively low number of choices of other domestic cities for a low intra-city decision city. Finally, Porto Alegre and Salvador combine high intra-city choice with high foreign choice, a corollary of the few linkages with other Brazilian cities referred to previously. These are the only two decision cities that use no Sao Paulo services (Table 3).

Towards an explanation of the transactions geography

Tables 3, 4 and 5 show the geography of the 218 transactions links but to begin to explain these patterns we have to draw upon the information we have on the characteristics of the firms creating production chains. The frequencies of transaction types reported in Table 4 can be disaggregated in four ways: by firm ownership, the decision-making firms’ sector, the scale of business, and the sector of service being outsourced. Analyses using these breakdowns of the data are carried out into two stages. First, we discuss whether the potential explanatory variables (data divisions) have a statistically significant effect on the transaction choices (see Appendix 1-4). Simple chi-square tests are used for this task. Second, where a significant statistical effect is found, we look for how the variable’s influence is featured through transactions emanating from the decision cities.

The chi-square tests are presented in the appendix; their results are very clear.

• Ownership of decision-making firms has a very strong influence on location decisions for outsourcing transactions (Appendix 1 refers to Tables 4 and 6).
• The sectors to which the decision-making firms belong have a very strong influence on location decisions for outsourcing transactions (Appendix 2 refers to Tables 4 and 7).
• The scale of the business, whether Brazilian or international, has a very strong influence on the location decisions for outsourcing transactions (Appendix 3 refers to Tables 4 and 8).
• The service sector being outsourced has no discernible influence on the location decisions for outsourcing transactions (Appendix 4).

These initial findings point to an understanding of the geography of the transactions. Three important influences on the location decisions for outsourcing have been discovered. The fourth possible explanation provides a negative statistical result but is nonetheless an interesting result. We can find no statistical differences in the outsourcing of different services. This means that we will not proceed to the second stage of the analysis with this variable: differences between services will not be considered further. For the other three explanations, we look at transaction frequencies across the decision cities to develop further understanding of the processes in the location decisions.

Table 6 shows the distribution of transaction for decision cities in terms of the ownership of the firms doing the outsourcing. The frequencies tested by chi-square in the appendix are the penultimate row of totals; in the last row these are presented as percentages within each ownership category to aid interpretation. Thus we can see that both foreign owned firms and Brazilian private firms overwhelmingly feature intra-city choice decisions (over 70%). This is largely due to the dominance of Sao Paulo and Rio de Janeiro as decision cities and service cities. Note, however that Porto Alegre also contributes to this result. The Brazilian state sector and mixed ownership firms differ from this pattern in different ways. The state-owned firms have a clear majority (59%) of its transactions going to other Brazilian cities other than to the one in which the decision is being made. Here we see Belo Horizonte and Curitiba’s contributions to the configuration of transactions. This ‘national distribution’ pattern contrasts with the mixed firms who spread their transactions evenly across the choice categories. This means that they have by far the highest frequency for foreign choice of servicing – the contributions of Belo Horizonte and Salvador are to be found here.

One question in particular arises from the above analysis: why do firms under mixed domestic and foreign ownership make much more use of foreign services than foreign-owned firms? First thoughts would suggest the latter should have at least as many, but in fact foreign servicing is the lowest percentage of foreign firm’s service choices. This is not a conundrum if we think of the differences between the how the two ownership categories locate their headquarters. Foreign firms locate their headquarters mostly in Sao Paulo and Rio de Janeiro and part of the reason for the decision is the quality of the services available in these two world cities: hence the high number of intra-city choice decisions. Mixed firms, on the other hand, are the result of mergers, which means that the headquarter decision is not made: there is already a Brazilian headquarter. In this case, the decision cities are not themselves the major service cities, and the foreign ownership link provides opportunities to use foreign as well as Brazilian-based services.

Table 7 shows the distribution of transactions by firm sectors arrayed across the decision cities. The critical feature of this Table is that all innovation firms’ transactions emanate from Sao Paulo and they are all intra-city decisions. There is also a contrast between the traditional and basic services with the former having appreciably more intra-city choice and the latter more choices in other Brazilian cities. This is a feature of the traditional industries being located in the main service cities of Sao Paulo and Rio de Janeiro (and also Porto Alegre) where intra-city choice is typical, as we have previously noted. The basic sector is located outside these service cities and therefore decisions to use services will not, to a large degree, be local city ones.

Table 8 shows how transaction decisions are distributed across cities in terms of whether the business is domestic or foreign. The fact that this variable is important is the least surprising of all our findings: there were no foreign transactions for servicing domestic business. However, the starkness of this result provides a pause for thought about the lack of using services outside Brazil when it comes to business that is not transnational. This does not mean domestic business has avoided being penetrated by global service firms; quite the opposite, there are very many global service firms in Sao Paulo and Rio de Janeiro: for instance, all outsourcing in accountancy is provided by global firms. Thus what this lack of the use of non-Brazilian service cities for domestic business indicates is that the large Brazilian cities are integral to the world city network created by such firms therefore abrogating the need to use foreign service cities. There are also differences between intra-city choice and other domestic choice with servicing foreign business more likely to be carried out as intra-city choice in Sao Paulo and Rio de Janeiro whereas servicing domestic business is more likely to be carried out in other Brazilian cities, notably Belo Horizonte and Campinas.

Transaction links and the gateway city thesis

Now that we have shown that there are systematic processes operating in Brazil to influence outsourcing of advanced producer services, we can consider the degree to which these transaction links support the idea of a ‘gateway city’ linking Brazil into the world economy under conditions of contemporary globalization. Are there gateway city processes operating and how strong are they? To answer this question we have to focus on how Sao Paulo fares in our analyses. This city is generally accepted as Brazil’s world or global city (Taylor 2004) and its gateway credentials have been argued by John Friedmann (1995): Sao Paulo is only one of only two cities identified as articulating national economies from ‘semi-periphery countries’ into the world economy. However, Rio de Janeiro is also sometimes identified as a world city (Taylor 2004, 70-1) and therefore this city needs also to be kept in mind. Furthermore, Tolosa (2003) has identified a ‘Rio/Sao Paulo Extended Metropolitan Region’ as a potential multi-nodal ‘global-city region’ (Scott 2001) and consequently we need to be mindful of this possibility as a ‘gateway city-region’ for Brazil.

The following empirical evidence supports the role of Sao Paulo as an active gateway city.

• There is no penetration into domestic business by services provided outside the country. Confirming the inheritance of a strongly urban-based economy, this shows the lack of service dependence, a precondition for a service gateway city.
• Firms in Brazil outsource twice as many services to Sao Paulo as to all foreign cities; Rio de Janeiro also attracts more transactions than combined foreign cities, This confirms Sao Paulo, and to a lesser extent Rio de Janeiro, have the capacity for endogenous advanced producer servicing of the Brazilian economy in a globalizing world economy
• Overall firms in Brazil outsource services to Sao Paulo offices twice as much to the second most popular city ( Rio), with nearly a 6:1 ratio to the third ranked city. A combined Sao-Rio region has approximately a 15:1 ratio in outsourced services in relation to next city. These represent high levels of economic concentration.
• Sao Paulo has the highest level of intra-city transactions, and all but one of the few choices of firms in Sao Paulo to use ‘outside’ services are transactions to foreign cities. This confirms Sao Paulo’s supremacy as the advanced service centre within Brazil.
• Foreign-owned firms in Sao Paulo are the firms most likely to generate intra-city transactions (87%). This confirms service sufficiency, capacity and concentration in Sao Paulo.
• All the transactions by ‘innovative firms’ originate in Sao Paulo and all these transactions remain within the city. This is the cutting edge sector of globalization and Sao Paulo is at the centre of it in the Brazilian economy.

All these results provide a strong case for interpreting Sao Paulo as Brazil’s gateway city in the world economy. But this is only part of the story.

All empirical categories privilege some evidence at the expense of other evidence. This is inevitable in their simplification of a complex reality. The question, therefore, is not whether a concept such as gateway city is an accurate portrayal of a situation, but the degree to which it fits the situation. In this case we have previously suggested that the concentration processes leading to gateway functions are balanced by powerful dispersal processes in economic globalization. Thus the strong case for Sao Paulo’s gateway function should be matched, in part at least, by simultaneous by-passing the gateway. Here is the evidence marshalled for this side of the story.

• As well as Sao Paulo and Rio de Janeiro there are two clear cases of self-sufficient service cities: Porto Alegre and Salvador, both outside the central core region of Brazil, make little or no use of Sao Paulo’s services.
• Belo Horizonte , Brazil’s other leading city, is not self-sufficient but has approximately the same number of transactions with foreign cities (9) as with Sao Paulo (10).
• Joint Brazilian/foreign owned firms, in particular, use foreign cities for servicing their production.
• Nearly half of foreign business servicing (47%) is carried out in foreign cities.
• Overall foreign cities account for 33 out of 218 transaction links – contra to the gateway case, this 15% of transactions can be deemed high for such a large country with numerous big cities.

Thus there is a second plausible position to be gleaned from our evidence that takes us beyond a simple gateway function.

Concluding comments

This paper has contributed three additions to the world cities and economic globalization literatures. First, at a conceptual level, we have combined the concentration/dispersion nexus of globalization studies with the gateway city concept of urban studies. Thus we have produced a new reading of the gateway process as part of a more complex pattern of globalization. Second, at a measurement level, we have created unique data for production-service transactions based upon cities. Thus we have been able to begin the task of studying the world city network enveloping production chains in new forms of analysis. Third, at an empirical level, we have contributed to putting more non-core cities ‘on the map’ of world cities through a detailed study of Brazilian cities in globalization from these new perspectives. Thus we have new evidence for how one major non-core country is integrating into a globalizing world economy.

Since this is essentially an empirical paper we shall conclude by elaborating on the latter contribution. Sao Paulo came the prominence as Brazil’s dominant city before contemporary globalization; gateway function processes within globalization have enabled the city to continue its dominance. This is true also for the maintenance of the second city status for Rio de Janeiro. Thus Sao Paulo and Rio de Janeiro easily outrank all other Brazilian cities as both service and decision cities. They appear to be holding their own as cities ‘articulating’ the Brazilian economy into the encompassing world economy for both domestic and foreign-owned firms. However, there are globalising processes that are by-passing the gateway. Firms in Porto Alegre and Salvador are in no sense dependent on Sao Paulo or Rio de Janeiro for their production-service transactions. Thus, as well as concentration in Brazil’s core-region, these cities from other coastal regions imply the growth of a broader littoralization of production-service links in the current phase of the Brazilian economy. In addition there are foreign servicing links that by-pass Sao Paulo and Rio de Janeiro that derive from, but not only from, Belo Horizonte. In all these processes Brasilia is largely inactive suggesting that firms in the capital city are of little relevance in Brazil’s economic globalization. Clearly, what we have shown in that Brazil appears to be a textbook case of the new economic complexity wherein global concentration and dispersion of economic activities are occurring simultaneously. This conclusion brings together current contrary interpretations of the Brazilian economy that champion just one of either increasing concentration or new dispersion. Both are happening.

REFERENCES

Aguilar, A D and Escamilla, I (eds) (1999) Problems of Megacities. Mexico City: Universidad Nacional Autonoma de Mexico.

Andersson, A. E. and Andersson, D. E., (eds) (2000), Gateways to the Global Economy. Cheltenham, UK: Edward Elgar.

Brown, E, Derudder, B, Panreiter, C, Pelupessy, W, Taylor, P J, Witlox, F (2006) Spatialities of globalization: towards an integration of research on world city networks and global commodity chains, GaWC Research Bulletin 148

Burghardt, A. F. (1971), A hypothesis about gateway cities. Annals of the Association of American Geographers, 61, 2, pp. 269-285.

Dicken, P. (2003) Global Shift: Transforming the World Economy, London: Paul Chapman.

Drennan, M. P. (1992), Gateway cities: the metropolitan sources of US producers service exports. Urban Studies, 29, 2, pp. 217-235.

Exame Melhores e Maiores (2003), Sao Paulo: Abril.

Fernandes, A.C. and Negreiros, R. (2001) ‘Economic developmentism and change within the Brazilian urban system’, Geoforum, 32, 415-435.

Fernandes, E. and Valença, M.M. (2001) ‘Urban Brazil: past and future’, Geoforum, 32, No 4, v-viii.

Friedmann, J. (1986) ‘The world city hypothesis’, Development and Change, 17, 69-83.

Friedmann, J. (1995) ‘’Where we stand: a decade of world city research’, in P. L. Knox & P. J. Taylor (eds) World Cities in a World-System, Cambridge: Cambridge University Press.

Gereffi, G. and Korzeniewicz, M. (eds) (1994) Commodity Chains and Global Capitalism . Westport: Praeger.

Grant, R. (2001), Liberalization polices and foreign companies in Accra, Ghana. Environment and Planning A, 33, pp. 997-1014.

Grant, R. and Nijman (2002) ‘Globalization and the Corporate Geography of cities in the less-developed world’, Annals of the Association of American Geographers, 92, No 2, 320-340.

Gugler, J (ed) (2004) World Cities beyond the West. Cambridge: Cambridge University Press.

IPEA, IBGE and UNICAMP (2001) Caracterização e tendências da rede urbana do Brasil, Brasilia: IPEA.

Johnston, R. J. (1982) The American Urban System, Longman, London.

Kipnis, B. A. (2004) ‘Tel Aviv, Israel – a world city in evolution: urban development at a deadend of global economy’, in M. Pak (ed), Cities in Transition, Ljubljana: Department of Geography, University of Ljubljana, 183-194.

Lo, F-c and Yeung, Y-M (eds) (1998) Globalization and the World of Large Cities. Tokyo: United Nations University Press.

Lyons , D and Salmon, S (1995) ‘World cities, multinational corporations, and urban hierarchy’ in P L Knox and P J Taylor (eds) World Cities in a World-System. Cambridge: Cambridge University Press.

Olds, K and Yeung, H (2004) Pathways to global city formation: a view from the developmental city-state of Singapore,Review of International Political Economy, 11 (3), 489-521.

Pelupessy, W. (2001) Industrialization in global commodity chains emanating from Latin America, UNISA Latin American Report 17(2), 4-14.

Pred, A (1977) City-systems in Advanced Economies. London: Hutchinson.

Robinson, J (2002) ‘Global and world cities: a view from off the map’, International Journal of Urban and Regional Research, 26, 531-54.

Robinson, J (2005) ‘Urban geography: world cities, or a world of cities’, Progress in Human Geography, 29, 757-65.

Rossi, E. and P. J. Taylor (2005) ‘Banking networks across Brazilian cities: interlocking cities within and beyond Brazil’, Cities, 22, No 5, 381-393.

Rossi, E. and P. J. Taylor (2006) ‘Gateway cities in globalization: how banks are using Brazilian cities’, Tidjschrift voor Economische en Social Geografie. 97, in press.

Santos , M. (1993) A Urbanização Brasileira, São Paulo: Hucitec.

Sassen, S. (1991/2001) The Global City, Princeton, NJ: Princeton University Press.

Sassen (1994) Cities in a World Economy, Thousand Oaks: Pine Forge Press.

Scott, A. J. (2001) ‘Globalization and the rise of city-regions’, European Planning Studies, 9, No 7, 813-826.

Short, J. R., Breitbach, C., Buckman, S. and Essex, J. (2000) ‘From world cities to gateway cities’, City, 4, 317-40.

Taylor, P. J. (2004) World City Network: a Global Urban Analysis, London: Routledge.

Taylor . P J (2006) The rise of ‘global’ advertising from 1990, GaWC Research Bulletin 192

Tolosa, H. (2003) ‘The Rio/Sao Paulo Extended Metropolitan Region: A quest for global integration’, The Annals of Regional Science, 37, 479-500.

Ullman, E. L. (1957), American Commodity Flow: a Geographical Interpretation of Rail and Water Traffic Based on Principles of Spatial Interchange, Seattle, University of Washington Press.

NOTES

* Eliana C Rossi, Production Engineering, COPPE, Federal University of Rio de Janeiro, RIO DE JANEIRO, CEP 21945 970, BRAZIL, Jonathan V Beaverstock and Peter J Taylor (corresponding author), Department of Geography, Loughborough University LOUGHBOROUGH, LE11 3TU, UK

¹ Here we use an analogy previously employed to argue the importance of one particular professional service, advertising: ‘Using an ecological analogy, an ‘indicator species’ in an ecosystem is not necessarily the ‘dominant species’, this is defined as the species around which the ecosystem is organized. An example would be the tallest tree species in a forest as dominant species, with monitoring of an owl species used to judge the health of the forest ecosystem. Thus indicator species are defined as critical within the food chain but they are not the dominant species. Advertising is critical in the contemporary commodity chains of the world-economy but it is by no means the largest industry – for instance, its firms do not feature prominently in the Fortune 500 .’ (Taylor 2006, 8) This argument is generalised here to cover all advanced producer services in economic globalization.

² When the time the survey was carried out, the status of some of the firms listed on EXAME had changed due to merging/acquisitions: the initial number of 120 firms decreased to 119: TELESP Celular and Vivo-RJ belong now to VIVO, a joint-venture of Portugal Telecom and Telefónica Móviles

³ In some firms the interaction was blocked by one of the company departments in spite of acceptance on the part of the president to take part in the project; the reason given to justify such action was that the required information was private.

⁴ The printed edition of Exame Melhores e Maiores of 2003 is the source for firm classification by sector. The three sector categories (Traditional, Innovative and Basic) aggregate those sectors according to technology or research uses and competition for market share (Boter and Holmquist, 1996). Sectors whose firms use essentially an established technology and compete for market share were allocated to the Traditional sector. Sectors whose firms operate in essence developing new products or mediating research and end users and compete for market share were allocated to the Innovative sector. The sectors whose firms operate with an established technology and no competition were grouped together in the Basic sector.

Table 1: Population and sample frequencies by basic characteristics

* as defined by Exame Melhores e Maiores

Table 2: Sectors and ownership for firms and transaction links

Table 3: Distribution of transaction links between service cities and decision cities

Table 4: Location choices in decision cities

Table 5: A typology of decision cities based on choices

Table 6: Location choices in decision cities by firm ownership

BS = state-owned; BR = Brazilian private ownership; MX = joint Brazilian/foreign ownership; FO = foreign owned

Table 7: Location choices in decision cities by firm sector

TR = traditional sector; IN = innovative sector; BS = basic sector

Table 8: Location choices in decision cities by business locale

DB = domestic business; FB = foreign business

Figure 1: Production-service transactions through cities

Appendix: statistical analyses

1. Location choice by firm ownership

Chi-square = 40.07, degrees of freedom = 6
Threshold at probability of 0.001 = 22.46
Decision: there is a very strong relationship between choice and ownership

2. Location choice by firm sector

Chi-square = 21.95, degrees of freedom = 4
Threshold at probability of 0.001 = 18.46
Decision: there is a very strong relationship between choice and sector

3. Location choice by domestic/foreign business

Chi-square = 85.34, degrees of freedom = 2
Threshold at probability of 0.001 = 13.82
Decision: there is a very strong relationship between choice and business locale

4. Location decision by service sector

Chi-square = 6.35, degrees of freedom = 10
Threshold at probability of 0.05 = 18.31
Decision: there is no relationship between choice and service sector

5. Domestic Business – Firm and Services Ownership

6. Foreign Business – Firm and Services Ownership

Note: This Research Bulletin has been published in Geoforum, 38 (4), (2007), 628-642