From Sectors to Functions: Producer Services, Metropolization and Agglomeration Forces in the Ile-de-france Region

metropolisation and agglomeration forces in the Ile-de-France region" Abstract: This paper reviews the role of producer services in the scientific debate on metropolisation processes and discusses the different types of agglomeration economies. Based on the critical appraisal of sectoral based approaches, it develops a more functional case study of the geography of producer services in the Paris metropolitan region to test the form and nature of economic specialization. The results confirm the multipolar-monocentric pattern which seems to be a common feature in many European cities. The paper concludes on the importance of a functional and hierarchical division of labor in the Paris city-region. economies Résumé : Les services aux entreprises jouent un rôle important dans les débats sur les formes et les facteurs des processus de métropolisation ainsi que sur la nature des économies d'agglomération. La critique des approches purement sectorielles généralement utilisées dans la littérature invite à un travail d'analyse plus fonctionnel. La géographie des services aux entreprises dans la région francilienne est réévaluée à partir d'une lecture de ce type et montre l'intensité de la division fonctionnelle et hiérarchique des tâches. Ceci confirme non seulement l'existence d'un "modèle" multipolaire et mono-centrique dans le cas francilien, faisant écho à ce que l'on trouve dans de nombreuses autres régions urbaines européennes, mais aussi l'importance des processus de division fonctionnelle et hiérarchique du travail dans le cadre de l'avènement d'une économie de l'information.

Advanced Producer Services (APS) are often identified as critical actors exercising the practice of command and control functions in the contemporary economy (Sassen, 2001). They play a leading role in metropolisation processes both at inter-and intra-city-region levels (Bouliane et al., 1998, Daniels, 1991, Daniels et al. 1996, Derudder, 2006. At the inter-urban scale, APS professionals are central in the development of a "geography of flows" (Castells, 1996) that is responsible for the "archipelago" economy described in the 1990s (Veltz, 1996), an economy which links major global metropolitan areas in one or more networks (Taylor, 2003). Following the same line of argument at intra-city-region level, the changing spatial organisation of metropolitan areas has been linked to the role of Business Services (BS) and more specifically of APS to explain for the shift from traditionally mono-centric cities built around a central economic core towards more polycentric urban regions (Kloosterman and Musterd, 2001,). The deconcentration of some business services from the CBDs to their surrounding suburbs -as first observed in the 1980s and 1990s in North-American cityregions -has been central in the description of metropolitan regions' contemporary transformations.
However, Canadian researchers have offered a critical appraisal of CBD decline (Coffey and Polese, 1996) which has led to a widely acknowledged change in the perception of current metropolitan transformations. With the death of cities no longer being taken for granted, the literature insists on the very strength of metropolitan central areas not only in European but also in North-American cityregions. After all, some APS do not leave the traditional CBD, but even tend to be more concentrated in central areas as less advanced business services are forced to relocate into less expensive suburban office markets.
Furthermore, as the New Economic Geography has been evoked by economists and some geographers to explain for the changing spatial organisation of metropolitan areas (not only at the level of urban systems but also within urban regions), producer services have kept continued to be central to the understanding of the formation of the polycentric urban regions. Following the hypothesis of a post-industrial economy, locational choices of producer services are sometimes thought to be driven by the search to improve both market externalities (proximity to clients for instance) and specific sectoral economies resulting from the co-location with other similar firms (localisation externalities).
The polycentric region would then be the result of firms' sectoral locational choices. However, it can be argued that a more trans-sectoral shift has occurred with the development not so much of a post-industrial economy but of an information economy (Porat, 1977, Reich, 1991, which affects all sectors, both services and industries. Producer services would be emblematic not so much of firms' sectoral considerations but of a functional differentiation taking place within metropolitan regions (Duranton, Puga, 2000, Feser, 2003. The polycentric city-region would thus result from an increasing division of labour not between different business services but between different functions of the economy. In the context of an information economy where access to tacit and face-to-face contacts proves to be crucial for the efficiency of some high valueadded activities, the most information-sensitive jobs would tend to remain in economic cores while others would relocate in peripheral places (as seen in the Fujita and Ogawa model, 1982). The centripetal forces would thus rather be linked to Jacobs-type externalities (Jacobs, 1969) with the potential interactions resulting from the co-location of advanced functions of differentiated economic sectors explaining the concentration of industrial headquarters and of numerous advanced services producers in the CBDs while less information-intensive functions (often depicted as back-office activities dealing with routinised tasks) would be decentralised to more secondary economic poles in the peripheries (or sometimes outside the city-region).
Considering that producer services are crucial to the understanding of contemporary spatial and economic changes occurring in metropolitan regions, this paper further develops the study of the functional division of labour hypothesis by addressing some of the methodological difficulties encountered in research so far. Much research has studied the functional division of tasks within metropolitan regions with data resulting from sectoral classifications, as has been the case with most analyses of the economy of the Paris metropolitan region (Crague, 2002, Halbert, 2004a, Boiteux-Orain et al., 2006. In this paper, I aim not only to show how, in the case of the Paris metropolitan region, the geography of producer services has evolved towards a more multipolar geography in the last decades of the 20 th century, but also how it is the result of a functional differentiation rather than a "new" sectoral change as the post-industrial hypothesis tends to over-emphasize. In other words, even though I do recognise the existence of a sector-based differentiation of activities in the multipolar geography of the Paris region, I argue that functional and hierarchical criteria seem to be more important factors in an understanding of contemporary intra-metropolitan transformations. The paper will start with a review of the literature on the forms, nature and factors of the changing intra-metropolitan geography of producer services and on their impacts regarding the formation of multipolar city-regions (1). I will then develop a study of the economic geography of the Paris region. After showing the "concentrated deconcentration" (Hall, Pain, 2006) occurring in the dense part of the Paris metropolitan region (2), the paper engages a functional analysis of producer services which goes past what appears to be a limited sectoral approach. Based on statistical analysis from the National Census Bureau's (Insee) database, a functional and hierarchical typology is proposed (3). The paper concludes with a discussion of the various externalities offered by metropolitan regions, analysis of the dynamics explaining the transformation of metropolitan regions and suggestions for some factors that should be emphasised in future research (4). 1) From the "death of cities" to CBD reinforcement: theoretical and methodological issues Metropolisation is often described as a set of two-fold selective dynamics: the concentration of conception, command and coordination functions in major urban areas and their parallel deconcentration within these urban areas. The latter process has given birth form the 1990s onwards to at least three major geographical debates on the pattern, nature and models of formation of contemporary metropolitan spaces in which producer services have proved crucial.
Business services in general, and Advanced Producer Services in particular, have long been associated with the central areas of cities. Whereas throughout the 19 th and much of the 20 th century most large cities like London, 6 New York or Paris faced intense absolute or relative deconcentration of their population and manufacturing activities, producer services professionals remained located in Central Business Districts. However since the 1960s (Abler, 1970), metropolitan productive systems have undergone a major change with the deconcentration of producer services not only from central places to newly developing regions (the rise of the "Sun belt" is an example in the U.S.) but also from traditional economic cores to intra-metropolitan peripheries. After the first two waves of deconcentration affecting population (and induced services to households) and manufacturing activities, scholars acknowledged a "third wave" (Cervero, 1989), and a "new suburbanization" (Stanback, 1991) that was thought -and is still considered by some -to dramatically change the nature of cities. In summary, during the Fordist era, centrality was a notion that could be  Glaeser, 1998). For example, in 1968, Webber saw the coming of the "post-city age" while in 1995, Gilder depicted cities as being "the leftover baggage from the industrial era". The extending suburbs of Los Angeles are cited as evidence of a "post-industrial and post-modern city model" (Muller, 1981(Muller, , 1997. To some analysts of the 1990s, urban sprawl was not a counterproductive anomaly that increased transaction costs, but the desirable future of urban spaces (Gordon, Richardson, 1996a, 1996b. The very forms of this new suburbanization have been discussed among North-American observers. For some, the deconcentration of producer services was seen as anarchical and dispersed (see Gordon, Richardson, 1996a and1996b and more recently Lang, 2003). At most, some large sub-regional areas may benefit more from this dynamic than the rest of the metropolitan area (high-tech quadrants like Silicon Valley, Route 128 or Orange Country). The resulting spatial organisation is one of dispersal and even spatial dilution of producer services. Alternatively, some authors observed what has been called in the European context a "concentrated deconcentration" (Hall, Pain, 2006).
Rather than dispersal, polarisation would prevail with the emergence of secondary centres that concentrate suburban offices (Archer, Smith, 1993) in "magnet areas" (Stanback, 1991), "Suburban Employment Centres" (Cervero, 1989) or "edge cities" (Garreau, 1991). However, if the spatial patterns differ considerably (multipolar vs. dispersed city-region), in both cases the logic remains the same. The post-modern city sees its economic "heart" shift to the outskirts of the city-region (Scott et al., 2001). Compared to the long inherited cities embodied by the old European model, the post-Fordist American metropolises were depicted as growing "inside out" (Soja, 2000), leading the way to a new and somewhat universal urban model of which Los Angeles was thought to be the norm rather than an outlier (Gordon, Richardson, 1996b).
However this interpretation of the nature of the deconcentration of business services has been controversial and was challenged by another proposal which has gained credit not only in Europe but in North America as well. If most authors agreed on the necessity to go past a centre-periphery analysis because of the dramatically new nature of urban and economic geography in metropolitan areas, it was still a common feature at that period to develop case studies dealing with only three or four types of spaces for the entire urban region: the CBD (or even the central city) and two or three surrounding rings (see for instance Gordon, Richardson, 1996a). Such broad perimeters made it very difficult to evaluate the form of deconcentration processes (i.e. polarised vs. terms of advanced producer services, almost all secondary poles emerging in the peripheries were defined by the total number of jobs and of office spaces (see Garreau, 1991), regardless of the nature of the jobs. There is hardly any attempt to identify the economic sectors, functions and/or socio-professional categories of the activities being deconcentrated from the central area. It is therefore impossible to qualify the nature of the deconcentration process, except by some very fragmentary interviews. How then can one conclude that advanced services' firms are migrating to suburbs (Garreau, 1991)? One may reverse the proposal and see in the deconcentration process the selective concentration of advanced services in CBDs resulting from the relocation of low-skilled and low value-added business services outside the city. Paradoxically, far from being a sign of the decline of the centre, this would indirectly be an effect of its strength.
This is more or less what some Canadian geographers argued when discussing the so-called edge cities model (Coffey and Polèse, 1996). Rather than focusing on peripheral places and their spectacular growth in terms of employment or office spaces, these authors studied the characteristics of the firms remaining in the Montreal and Toronto CBDs. They observe an intensified specialisation in high value-added and international producer services. The CBDs remains -or even becomes -increasingly central for high level functions of the productive system.
Empirical studies showed that two types of firms were -and one would argue are still -concentrating in the CBDs: first, the face-to-face-intensive firms dealing with coordination and management activities; second, large industrial, service and commercial companies' headquarters that benefit from the concentration of business services producers. These two groups are in permanent interaction: numerous meetings, informal and formal relationships attest to a business environment that has some characteristics of Marshallian or industrial districts.
The firms leaving the central area of the metropolitan regions are of a different nature: they are the final demand-oriented businesses dealing with customers and/or the low value-added routine task workers (call centres, back-offices, etc.). Thus the necessity to carefully define the content of an often too evasive 'advanced services' category. externalities. In this latter explanation, two distinct -but potentially complementary -economic logics are analysed. The sectoral approach that has been used in most multipolar models (Fujita and Ogawa, 1982) relies on localisation externalities: agglomeration forces result from the higher economic profitability found by firms in locating close to identical firms. In a one-sector economy (in our case, the producer services sector), the concentrated deconcentration leading to a multipolar system would result from the necessity of producer services to be located together -for instance to exchange tacit information -while the transport costs of their specialised workforce act as dispersion forces which, at a certain level, induce the development of secondary economic poles to minimise commuting costs. In a two-sector economy (industry and producer services for example), the multipolar geography might be constituted with one or more poles specialised in producer services and in a series of other secondary industrial poles: this results from differentiated sensitivity to localisation externalities between the two sectors.
However, this first sectoral approach is also balanced in another Fujita and Ogawa model by a functional organisation (1993). Agglomeration forces would result from Jacobs-type urbanisation externalities (Jacobs, 1969): "front-office" activities that are intensive in tacit information exchanges concentrate in the core of a given urban region while less information-intensive "back-office" activities can be located in peripheral areas to cut down location costs. The  14 transcription work, etc.). Moreover, this sectoral approach tends to rule out some workers involved in lower-order business services whose everyday tasks are of advanced level (managers, engineers, designers that work in more basic business services firms). Last but not least, this type of approach tends to keep the analysis within the realm of a sectoral understanding of the production system, thus diminishing our ability to directly observe the functional differentiation occurring in a given metropolitan economy. For example, if one uses legal services, accounting or advertising firms to stand for the category of advanced producer services, it is difficult to determine whether their intrametropolitan locations follow specific sector-based spatial logics or reflect a more transectoral functional organisation of advanced economic activities in the metropolitan area.
A second approach would try to differentiate producer services not by selecting high-order sectors that are supposed to represent advanced services but by looking at each producer services workers' everyday practice. In this analysis, the advanced producer services category would be made of all service producer workers in charge of conception, control and coordination functions in their firm, no matter the economic sector of their firms, whether it is logistic, IT or financial services.
The practical limit to this methodology is obviously the difficulty to access relevant data. In the French case, thanks to the updated functional analytical grid developed by the STRATES team in the 1990s, it is possible to use the detailed socio-professional dataset produced by the French Bureau of the Census (Insee). This database details workers' main activities in a classification of around 400 types of jobs. The data collection is done at each national census with a survey rate of a quarter of the active population thus allowing analyses at municipal level.
I have re-codified Insee's socio-professional classification into 14 major functions that constitute the production system (see Box 1). Although the names of the functions reflect a categorisation of the economy which tends to oppose "production" activities and more "collective" and often traditionally state-led functions (health, education, public administration), this analytical grid remains a powerful tool to improve our understanding of the changes occurring in the functional and hierarchical geography of business services in the Paris region. It is this analysis which I now undertake.
Box 1: A functional tool to analyse the production system  cultural activities but also with basic producer servicing jobs (catering, hotels); second, conception and marketing, but also to some extent public administration and education jobs are often found in the same places; third, material production counting for 8 % of the municipalities (Ile-de-France administrative boundaries) but for half of the regional employment and for more than 70 % of producer services jobs.      These selected example tend to enlarge the scope of most spatial economy models when addressing the mechanism for formation of metropolitan spaces (which are mostly based on firms' and workers' locational decisions) to take into account other actors as proposed in other works (Henderson and Mitra, 1996, Zang and Komei, 1997, 2000, Crouzet, 2003. I argue for instance that detailed analyses of real estate developers' strategies (promoters, investors, etc.) are still to be undertaken to develop a more encompassing theory of the production of metropolitan spaces. As firms outsource their office park, the role of real estate developers is becoming more important. I make the hypothesis that they tend not only to (partly) follow their clients' locational logics but also to superimpose their own objectives which are considerably dependent on structural changes happening in this particular industry (financialisation and globalisation).

<< insert Table 1 (Correlation Matrix) >>
All in all, advanced producer services, although important to understand the way structural economic changes affect city-regions, require to contextualised in a broader systemic framework within which a larger set of actors can be studied in terms of their activities and of their interactions.