Since the end of the 1990s, there has been a broad consensus regarding observations of an accentuation of social contrasts in the Paris metropolitan area, to the point that extreme terms such as “segregation” and “ghetto” are now frequently used in the public arena. Scientific debate, on the other hand, has focused on the theory of the “global city” and its corollary, the social and spatial polarisation of the metropolitan space (Sassen 1991), which draws a simplistic contrast between the beaux quartiers (wealthy districts), supposedly becoming ever more bourgeois, and neighbourhoods “in difficulty” that are apparently tending to become not only more homogeneous but also ever poorer (Maurin 2004). And yet, though such trends can be observed and indeed are highly visible, they do not tell the whole story, nor are they an inevitability. Other studies, for instance, have shown that the Paris metropolitan area appears to be moving away from this dichotomous model of the city, owing to the fact that, outside of these two extremes, the majority of private residential areas in the outer suburbs still appear to be inhabited mostly by middle-income households employed in non-managerial or manual jobs (Préteceille 2006).
How can we move beyond the apparent contradictions of these observations and consequently understand the dynamics, with all their territorial subtleties, that today underlie the complex social restructuring occurring in the Paris metropolitan area? Using exhaustive, localised social data, the aim is to identify the trajectories taken over time by different types of territories and by the spatial arrangement of these territories. In recent years, studies of this kind have been conducted in the Paris region, but, with a few notable exceptions (François et al. 2003), most of these use census data and therefore are concerned with socio-professional categories (Rhein 1998; Préteceille 2003). Our study  is based on a different choice, namely examining household income data for the Paris region.  This income-based approach was selected not only because it is generally complementary to approaches based on socio-professional categories (Chauvel and Chenu 2002), but also because of its propensity to provide comprehensive summaries. 
Our observations concern the period between 1990 and 2007, with particular attention paid to the years from 1999 to 2007. A number of key points define the regional framework of overarching dynamics considered here. Generally speaking, the median income of households in the Paris region (which was already higher than in the rest of mainland France in 1990) has increased less rapidly since the start of the 2000s than in other regions (Aerts and Chirazi 2010). At the same time, the gaps between income levels in the Paris region have widened. Furthermore, unlike the trends observed in other regions, the lowest incomes have increased less rapidly than the highest incomes. Observations of changes in income per decile also show that the higher the income bracket, the greater the increase in income.
Although these very general trends define the broad lines of the changes under consideration here, they tell us nothing about the way these changes relate to the region’s spatial dynamics. And yet references to these dynamics are omnipresent in the public debates on the future of this major metropolitan region  (Schéma directeur de la région Île-de-France (regional master plan), the creation of a “Greater Paris” council, etc.). Here, we have chosen to draw attention to the territorial characteristics of the social dynamics at play, taking as our starting point the observation of inequalities in household income at the level of individual communes (municipalities). This allows us to identify both the social disparities present at local level and the spatial trends behind these disparities across the whole region (François et al. 2011). The municipal level also has the advantage of forming a relevant framework for various aspects of residents’ day-to-day life, as well as for political action, in view of the range of responsibilities assigned to communes in France. These observations will be followed by inframunicipal investigations at the neighbourhood level.
Gentrification and accentuation of social polarisation
Two spatial trends have accompanied the changes in household income in municipalities in the Paris region since 1990  (figure 1). The first corresponds to a process of gentrification, which has been apparent continuously since 1990 and subject to a dual movement of retraction and diffusion. The number of municipalities identified as being home to “very well-off” (Type 1) and “well-off” (Type 2) households has fallen,  affecting smaller urban and rural municipalities first and foremost. Over the same period, the number of “fairly well-off” municipalities (Type 3) has, on the other hand, increased spectacularly,  to the point that they now represent over a third of all municipalities in the region, including a very large number of periurban towns and villages in the départements of Yvelines, Val-d’Oise and Essonne.
The second trend, just as prevalent but less unexpected, has involved an accentuation of social polarisation around two geographically clustered extremes, to the west and to the north of Paris respectively. These clusters have become increasingly opposed socially, while growing ever closer geographically. Specifically, despite a relative retraction in spatial terms, the residential cluster specialised in “well-off” and “very well-off” households, based in the west, reinforced its specificities between 1999 and 2007 (Types 1 and 2), while accentuating its spatial compactness, particularly in the urban core. In parallel, the second area, based in the north, also saw the specificity of its municipal income profiles increase, but towards greater relative poverty (Types 6 and 7). Between 1999 and 2007, this second area did not grow in size, but became more compact; municipalities belonging to these income profiles became increasingly concentrated in Seine-Saint-Denis, while the other key poor areas of the region (Mantes-la-Jolie and Trappes in Yvelines; the Seine valley in Val-de-Marne; the eastern edges of Seine-et-Marne) have shrunk considerably since 1999, or have even disappeared in some cases.
This accentuation of the specificity of extreme socio-spatial clusters is all the more significant given that, overall (at municipal level across the whole region), the links between social structures and income structures were reinforced between 1999 and 2007. For example, the social singularity and homogeneity of the cluster of “very poor” municipalities were accentuated with regard to particularly discriminating indicators relating to unemployment, lack of qualifications, and proportions of non-managerial workers, manual workers and non-nationals. This phenomenon was mirrored in municipalities defined increasingly by their relative concentration of “well-off” and “very well-off” households, which also saw their social singularity increase, particularly with regard to the proportion of residents in managerial jobs.
Homogenisation and growth of discontinuities
Significant spatial realignments have therefore been observed within the region. How have changes in socio-residential diversity in individual neighbourhoods contributed to these realignments? Some answers to this question can be found by considering the inframunicipal level, and more specifically the “IRIS” neighbourhoods defined by Insee, the French statistics office (figure 2).  First, it is increasingly rare to find neighbourhoods with profiles at opposite extremes within the same municipality. This mutual spatial exclusion of the most socially specialised areas, already significant in 1999, was further reinforced in 2007. The fabric of neighbourhoods in “very well-off” or “very poor” municipalities is tending to become more homogeneous,  while there has been a noticeable reduction in the total number of “mixed” neighbourhoods (Type 5), where the income profile is close to the regional profile, and of “poor” neighbourhoods (Type 6). This mutual exclusion is accompanied by a certain homogenisation of areas that, in 1999, exhibited a greater diversity of intermediate-type neighbourhoods and which acted as buffer zones between very socially specialised areas. This is the case, for example, in suburban areas in the east of Seine-Saint-Denis, or on the border between Val-d’Oise and Yvelines (in towns such as Ermont, Cormeilles-en-Parisis and Houilles), where a previously heterogeneous mosaic of neighbourhoods of various types is now dominated by neighbourhoods with “middle-income” (Type 4) and, above all, “fairly well-off” (Type 3) income profiles. The same applies to the city of Paris, where the beaux quartiers in the west and centre have extended eastwards, to the detriment of previously mixed areas (such as the 9th and 13th arrondissements), and where the remaining socially mixed areas in the north-east of the city have expanded, but at the expense of some of the last remaining poor neighbourhoods.
In parallel, the transitions between key zones of the Paris region, observed at the inframunicipal level, reveals greater discontinuities in 2007 than in 1999. Such cases include the town of Colombes, close to the business district of La Défense, or, to a lesser extent, the towns of Aulnay-sous-Bois and Sevran in Seine-Saint-Denis, where the income gap between low-rise residential areas and high-rise social housing estates has widened. This increase in relatively brutal contact between areas of contrasting income profiles is principally linked to the presence of islands of poverty. It is true that such islands were rarer in 2007 than in 1999, and that certain neighbourhoods – such as La Butte Rouge in Châtenay-Malabry or Les Chênes in Ermont – have disappeared from this category altogether. However, where these islands of poverty still exist (generally the larger and more populated areas, e.g. in Chanteloup-les-Vignes, Clichy-sous-Bois/Montfermeil or Bagneux), they typically contrast more violently with their immediate surroundings, which, conversely, are subject to gentrification and homogenisation. Elsewhere, the gentrification process, highly visible at inframunicipal level, can temporarily blur pre-existing local discontinuities. If we set aside these exceptional cases, the picture at the end of the 2000s is one of much less socio-residential diversity between adjacent neighbourhoods than 10 years previously, leading more often to significant spatial discontinuities, which underpin the spatial restructuring identified at municipal level.
The extremes in terms of income profiles are, fairly logically, associated with very low levels of inframunicipal diversity, as in Neuilly-sur-Seine (labelled 6 on the map above) or Gennevilliers (5). In the former, there is practically zero diversity; in the latter, however, a degree of diversity is suggested, with two neighbourhoods that have more of a “middle-income” profile. This illustrates a regional trend, namely a reduced tolerance of internal diversity in “very well-off” municipalities. Towns in less extreme situations (“fairly well-off” and “mixed”), such as Puteaux (7) and Courbevoie (3), sometimes offer only minimal internal diversity: their various neighbourhoods belong to the same municipal type. Typically, though, these municipalities outside the extremes cover a wide variety of neighbourhoods, as is the case in Colombes (2), La Garenne-Colombes (4) and Asnières-sur-Seine (1), where a social mix is clearly ingrained in the diversity of their neighbourhoods. These towns lie on the regional dividing line between the two extreme clusters.
Poor households pushed out to the (extraregional) periphery?
We can therefore conclude that there has been an overall reduction in socio-spatial diversity between 1999 and 2007, which denotes a dual movement of growing spatial contrasts and a simplification of the model for the social division of space in the Paris region. This simplification results from a more pronounced opposition between the two clusters of municipalities at the extremes, which have seen their social singularity (over-concentration of situations of considerable insecurity or considerable wealth), their homogeneity and their compactness all increase. Although we can therefore conclude that there has been an accentuation of the spatial polarisation in the Paris region, the dynamics observed do not result in a completely dichotomous metropolitan space. Indeed, a large majority of municipalities in intermediate positions between these two clusters have, over this period, been marked by gentrification processes under way since 1990.
At the same time, the regional model for 2007 appears less organised than in 1990, when clear patterns of income sectors were still evident. While such pattens remain very much present in the inner-ring départements, with an expansion of the “well-off” sector in the west and a retraction of the “poor” sector around a compact nucleus to the north of Paris, the same cannot be said of the outer-ring départements, where a homogenisation has been observed, with “fairly well-off” neighbourhoods extending outwards from the edges of the wealthiest areas – at the expense of mixed neighbourhoods – and islands of “poor” municipalities being gradually erased. 
This conclusion nevertheless raises a new question: does the geographical area chosen for this study mask the exclusion of low- and middle-income households, who can no longer afford to live in the Paris region? Many researchers and local stakeholders are today drawing attention to the diffusion of the region’s population (Nolorgues 2010; Beaufils and Louchart 2010), and in particular the arrival of poor households in areas just beyond the regional boundaries. Research into social inequalities and the way these inequalities change can no longer ignore these fringe areas that now lie on one of the dividing lines of the Paris metropolitan area, and where – in common with many parts of periurban France – a strong “sense of abandonment” (Davezies 2012) is being felt.
- Aerts, A. and Chirazi, S. 2010. “Les revenus des ménages entre 2002 et 2007. Un rééquilibrage entre territoires mais des disparités dans les pôles urbains”, Insee Première, no. 1309.
- Beaufils, S. and Louchart, P. 2010. “Creusement du déficit migratoire francilien, quels enseignements ?”, Cahiers de l’IAU, no. 153, pp. 21–23.
- Chauvel, L. and Chenu, A. 2002. “Enjeux et usages des catégories socioprofessionnelles : traditions nationales, comparaisons internationales et standardisation européenne”, interview with Marco Oberti and Edmond Préteceille, Sociétés contemporaines, nos. 45–46, pp. 157–186.
- Davezies, L. 2012. La crise qui vient. La nouvelle fracture territoriale, Paris: Seuil.
- François, J.-C., Mathian, H., Ribardière, A. and Saint-Julien, T. 2003. Les disparités des revenus des ménages franciliens en 1999 : approches communale et infra-communale et évolution des différenciations infra-communales 1990‑1999, Paris: UMR Géographie-cités/DREIA Île-de-France.
- François, J.-C., Fleury, A., Mathian, H., Pavard, A., Ribardière, A. and Saint-Julien, T. 2011. Les disparités de revenus des ménages franciliens. Analyse de l’évolution 1990‑1999‑2007, Paris: UMR Géographie-cités/DREIA Île-de-France.
- Maurin, É. 2004. Le ghetto français. Enquête sur le séparatisme social, Paris: Seuil.
- Nolorgues, F. 2010. “Quelle consommation d’espace dans les franges franciliennes. Le Bassin parisien, une méga-région ?”, Cahiers de l’IAU, no. 153, pp. 27–28.
- Préteceille, E. 2003. La division sociale de l’espace francilien. Typologie socio-professionnelle 1999 et transformations de l’espace résidentiel 1990‑99, Paris: Observatoire sociologique du changement (FNSP/CNRS).
- Préteceille, E. 2006. “La ségrégation sociale a-t-elle augmenté ? La métropole parisienne entre polarisation et mixité”, Sociétés contemporaines, no. 62, pp. 69–93.
- Rhein, C. 1998. “The working class, minorities and housing in Paris, the rise of fragmentations”, GeoJournal, vol. 46, no. 1, pp. 51–62.
- Sassen, S. 1991. The Global City. New York, London, Tokyo, Princeton: Princeton University Press.
 Data processing and cartography by Antonin Pavard, research engineer.
 In the form of “annual net taxable household income after deductions and alloxances, expressed per consumption unit within the household” (based on data from FILOCOM, the national file of dwellings at municipal level).
 It internalises the structural dimensions referred to by socio-professional categories, as well as effects linked to other vectors of social differentiation, such as those linked to employment situations (intermittent work, job insecurity) or more generally to all situations that fall outside the strict bounds of the occupational sphere (in particular the case of retired people). It thus takes into account internal shifts in social position, at a time when the content of professional categories is changing quite rapidly.
 The Paris region – Île-de-France in French – covers eight départements: Paris in the centre; an inner ring comprising (clockwise from the north) Seine-Saint-Denis, Val-de-Marne and Hauts-de-Seine; and an outer ring comprising (clockwise from the north) Val-d’Oise, Seine-et-Marne, Essonne and Yvelines. Although by far the most populated region of France, with 11.8 million residents, it is not entirely urbanised, with significant areas of countryside in the outer-ring départements. For this reason, the term “Paris region” is used in this article rather than “Greater Paris”, which refers more specifically to the urban area (and for which a new tier of local government is soon to be created).
 Disparities in household income are evaluated using a 10-band categorisation, corresponding to the income deciles in the Paris region. This categorisation is therefore specific to the region, justified by its specificities in terms of income levels and differences. In order to evaluate the change in multivariate states, the study seeks to position, via projection, the profile of each municipality in 1990 and 2007 on the 1999 reference index, i.e. a typology based on 7 types of municipality, taking into account both the income levels of the resident populations and the level of disparities. For more details on the method used, see François et al. (2011), pp. 32–33.
 The number of “very well-off” (Type 1) municipalities fell from 105 in 1990 to 46 in 1999 and to 41 in 2007; the number of “well-off” (Type 2) municipalities began to fall after 1999: 126 in 1990, 170 in 1999 and 159 in 2007.
 The number of Type 3 municipalities increased from 162 in 1990 to 460 in 2007.
 IRIS (îlots regroupés pour l’information statistique – grouped blocks for statistical information) areas were the second observation level selected. IRIS areas are the most detailed level at which Insee publishes inframunicipal data. For the method used to classify IRIS areas according to the types defined at municipal level, see François et al. (2011), pp. 55–57.
 For example, in 2007, Neuilly-sur-Seine was almost entirely composed of Type 1 neighbourhoods (with the exception of areas in the vicinity of Avenue de Paris, which were of Type 2); conversely, Villetaneuse and La Courneuve are now entirely composed of Type 7 neighbourhoods.
 This is a result of deindustrialisation and gentrification processes in the case of islands of poverty in the built-up area (in the Yvelines and Val-de-Marne départements), and linked to periurbanisation phenomena and the reduction of farming activities in the outlying parts of the region (Seine-et-Marne département).