The peri-urban as a statistical construction

10The classic definition of the city is based on the principle of a continuously constructed area. In the 1950s INSEE began by breaking down cities and their continuous suburbs into urban units or ZUUs (zonage en unités urbaines). Urban units are ranked by the number of jobs they contain, with urban “poles” (pôles urbains) distinguished from all other units (the initial breakoff point between an ordinary unit and a pole was 5,000 jobs). This classification was enriched in 1962 with the term “industrial and urban settlement zones” (ZPIU, zones de peuplement industriel et urbain), a notion that enabled statisticians to move beyond morphological criteria, which were deemed reductive, and to identify the areas within a given city’s “zone of influence.” ZPIUs were identified using a full set of socio-economic criteria: commuting to work, magnitude of the local farming population, population growth rate, and others. This category, used for 30 years to apprehend peri-urbanization, distinguished between urban units, rural municipalities, industrial municipalities, and municipalities commonly referred to as “dormitory” communities [1] (Berger 2004, Bretagnolle 2015). But since the criteria remained unchanged over that long period, the ZPIU classification lost its power to discriminate: in 1990, 96% of the French population were classified as living in ZPIUs, leaving only 4% in non-ZPIU rural areas. This led to overestimating the size of areas under city influence and therefore to overestimating peri-urbanization altogether (Le Jeannic 1996).

11So in 1996, ZPIUs were replaced by ZAUs—urban areas (zonage en aires urbaines)—an entity constructed on the basis of 1990 census data. This solved the problem. The ZAU classification was later modified several times to take account of peri-urbanization and to more effectively meet the needs of zoning users (Table 1).

12ZAUs or urban areas are determined on the basis of “urban units”: each urban area is made up of a continuous unit and a “peri-urban ring”, the latter made up of municipalities where at least 40% of the working population works in the related urban unit or in another municipality belonging to that urban unit’s field of attraction. This construction is based on an iterative process that takes into account rebound and snowball effects. For example, a municipality where 25% of the working population works in the urban unit and 20% in municipalities located in the unit’s periurban ring is classified as peri-urban (see Appendix, Map 1). This process results in an extensive definition of the peri-urban, whereas the 40% commuters break-off is more restrictive (Le Jeannic 1996). [2] In addition to rings of urban units, peri-urban areas encompass what are called “multipolarized” municipalities; that is, towns that are not part of the peri-urban ring but where at least 40% of working inhabitants work in at least two related urban areas even if the afore-cited break-off point is not attained for one of them. Peri-urban rings and multipolarized areas are defined by INSEE as continuous, i.e., without breaks or enclaves. Ultimately, INSEE defined urban areas in both economic and functional terms, combining the notions of employment catchment area and commuting area. The main purpose of the classification was to measure the influence of French cities on French territory as a whole (Brutel and Levy 2011).

13In 2002 INSEE further developed the 1996 urban area classification system by taking into account rural employment cores and their rings. The new zoning entity was called ZAUER (zonage en aires urbaines et aires d’emploi de l’espace rural), or urban area and rural employment area zoning, and it coexisted with the previous one. It was not until 2010 that a corner was turned with the new ZAU, established on the basis of 2008 population statistics (Appendix, Map 2). The main criteria (the 40% commuters threshold and the iterative process) did not change, but the new zoning was now used to rank urban areas (INSEE 2011): France’s 792 urban areas were now divided into 1) large urban areas, in which each unit accounts for over 10,000 jobs (230 such areas in European France); 2) medium urban areas, where the unit accounts for 5000–10,000 jobs; and 3) small urban areas, with a unit accounting for 1500–5000 jobs (Floch and Levy 2011).

14The effect of this hierarchical ranking of urban units and therefore of urban areas was to modify the perimeters of the peri-urban as measured at the national scale: raising the break-off point defining a large urban unit from 5000 to 10,000 jobs reduces the number of urban areas. Contrary to the previous system, peri-urban rings made up of units with 5000 to 10,000 jobs (i.e., “medium urban areas”) are no longer counted as peri-urban; the municipalities that compose those rings now fall into the “rural” category. Mechanically, then, this change reduced the urban area population by 6% and urban area employment capacity by 10% (Brutel and Levy 2011). Meanwhile, it reduced the amount of multipolarized suburban space by construction.

15As explained in the Cuillier Report (2009) of the National Council of Statistical Information or CNIS, the urban area category was restricted this way to strengthen size coherence between urban units and municipalities located in urban areas, the aim being to align statistical analysis more closely with public policy goals. The Cuillier report was used to shape the new zoning system. Meanwhile INSEE put forward an additional argument: “The break-off point of 10,000 jobs ensures that large urban units are consistent with people’s representation of urban-ness” (INSEE 2011: 10). Unless supported by further details, this argument is debatable, and INSEE has offered no evidence that the representation of urban-ness referred to is in fact shared.

16Another novelty of the 2010 zoning system concerns classifying municipalities as “urban” or “rural”: the distinction made by the 2002 ZAUER between “primarily urban space” and “primarily rural space” disappeared (Brutel and Levy 2011). “Primarily urban spaces” were either urban units or peri-urban areas, while “primarily rural spaces” included small urban units and rural municipalities that did not belong to “primarily urban space.” The CNIS was opposed to defining “rural” as the counterpart of “urban” in this way (Cuillier 2009). The new system now accounts for the co-presence of “rural” and “urban” municipalities in peri-urban space. The peri-urban ring of Paris, for example, encompasses slightly over 600 rural municipalities of under 2000 inhabitants alongside urban units, some of which are quite large. The urban unit of Meaux, with its six municipalities and 67,500 inhabitants, is the most heavily populated of them, followed by Fontainebleau (37,000 inhabitants), Étampes (27,500) and Coulommiers (24,500). Today these areas are an integral part of the Paris zone of influence (Julien 2000). This clarifies that peri-urban areas do not constitute a homogeneous whole. In addition to almost exclusively residential areas, they may include longstanding secondary centers that offer city-like public and commercial services. The presence of urban units within peri-urban areas also explains the substantial share of collective housing (apartment buildings) in them: in the peri-urban rings of Paris, Marseille-Aix-en-Provence, Nice and Strasbourg, nearly 25% of housing is collective (Cusin 2016).

Critical perspectives on the INSEE zoning system

17Few urban sociology studies of the peri-urban make intensive use of INSEE’s zoning system. Qualitative studies hardly use it at all, except to specify that the areas they have selected for study fit INSEE’s definition of the peri-urban. Quantitative studies, fewer in number and usually done by geographers, often use it “for want of anything better” and are quick to point out its limitations. Some researchers have developed their own ad hoc zoning systems, more relevant to their subjects of inquiry (Berger 2004, Roux and Vanier 2008, Wiel 2010).

18For Olivier Piron (2014), who has harshly criticized INSEE zoning classifications, an excessively inclusive definition of peri-urban space leads to an inflated vision of urban space—in the name of what he calls “urbanist imperialism.” As Piron sees it, INSEE’s announcement that 95% of the French population is living under urban influence today (Brutel and Levy 2011) amounts to an attempt to eradicate rural-ness from our representations of French territory. The author protests against replacing “rural units” with “small urban areas”—the result of instituting the new ZAU in 2010. He is also against an “economic” reading of French territory in terms of dependence on employment units, calling instead for a reading in terms of lifestyles and modes of “inhabiting”. While this last argument is admissible, it is probably important not to see the notion of “urban influence” as the reflection of a desire to eliminate the category of the rural. As we have seen, the new zoning system introduces the rural/urban distinction into what is now peri-urban space and—rightly or wrongly—its hierarchical ranking of urban “poles” reduces that space by construction.

19Emmanuel Roux and Martin Vanier (2008) also criticize the INSEE system for defining peri-urban space exclusively in terms of employment polarization and daily commuting. But they appreciate it for its simplicity, even though, as they point out, the ZAU classification cannot account for the diverse types of logic driving periurbanization processes. They explain that zoning classifications based on rebound effects result in overestimating peri-urban space whereas a dual zoning classification (i.e., using both the 40% and 20% break-off points) would identify municipalities undergoing peri-urbanization and so make it possible to predict development. They also claim that the INSEE zoning system forces us to see cities as having peripheral rings, belts or rims whereas maps based on new housing construction bring to light developments driven by other forces: corridor-shaped urbanization extending out from conurbations, or “micro-pigmentation” urbanization of large areas of rural space reflecting an entire set of local polarities. For Roux and Vanier, if we want to obtain an accurate view of change we need not just INSEE’s zoning system but inductive mapping of the advancing edge of built-up land.

20Critics also question the relevance of INSEE’s distinction between continuous city suburbs and the peri-urban ring. Here the argument is that INSEE zoning breaks up areas that are in some respects continuous: on either side of the “border” between suburbs and peri-urban zones we find areas with detached houses whose inhabitants’ socio-demographic profiles and lifestyles may be quite similar. To solve this problem and avoid artificially confining peri-urban expansion to INSEE-defined rings, Martine Berger (2004) studied the great ring around Paris, made up of four peripheral départements; [3] that ring encompasses remote suburban municipalities and rural Île-de-France municipalities that fall outside the Paris urban area. By proceeding this way, she is likely to have attained a more accurate view of the reality of periurban space in France in terms of location, type of habitat, and lifestyle. However, her breakdown by département cannot be transposed to other urban areas. And using it today would mean excluding from the analysis all non-Île-de-France areas now included in INSEE’s peri-urban and multipolarized spaces.

21Another point of disagreement concerns the common association between peri-urban development and urban sprawl. Theoretically, that association is highly debatable. Urban sprawl refers to a morphological urbanization process: an “oil stain” or “glove fingers” that push outward from conurbations. This type of urban growth began long before peri-urbanization; it formed the suburbs of French industrial cities in the nineteenth century and drove their considerable expansion throughout the twentieth. What is genuinely new in the urbanization cycle under way since the 1970s is not the dominance of centrifugal development but the type of that development. Peri-urbanization “eats away” at rural territories through the building of housing, facilities and premises for economic activities, not through sprawl in the classic sense of the term (Mayoux 1979). The centrifugal movement under way works to promote a “diffuse city” whose growth involves increasingly discontinuous, relatively sparse land use (Wiel 2010). The fact that real estate and housing tract developers are less and less implicated in detached house construction results in markedly discontinuous land development. And whatever sprawl there is does not start out exclusively from main urban units but also from small urban units and rural municipalities within peri-urban space. However this may be, INSEE’s ZAU is not well adapted to studying urban sprawl (Vanier and Roux 2008). But as we shall see, it is a useful tool for apprehending population growth in low-density spaces that are entering city zones of influence.

22We conclude this discussion of criticisms with two points that are crucial to understanding urban structures by way of the ZAU zoning classification. The first is the question of centrality. The ZAU apprehends space in accordance with the classic center-periphery opposition. In theory it allows for the existence of secondary, structuring centers (centers of other urban units located in the urban area), but the criteria nonetheless force upon us a representation of urban systems as having one center only (Cusin 2016). The fact is that urban areas like Marseille-Aix-en-Provence, Nice or Lille differ considerably from this simple model because they are multipolarized, in contrast to Lyon, Toulouse, Nantes or Rennes.

23The second point concerns the fundamentally dynamic nature of peri-urbanization and the ensuing methodological problems. Given the fast pace of development, it is important to update the ZAU classification regularly on the basis of new census data. The perimeter of peri-urban space changes continually, making time comparisons more difficult, with peri-urban municipalities or urban units that are becoming part of suburbs on one hand, and municipalities and urban units in the process of becoming peri-urban on the other. Another phenomenon to take into account is urban area fusion. For example, as INSEE was adopting the 2010 ZAU, based, as we know, on 2008 census data, the urban areas of Elbeuf, Forbach, Saint-Chamond, and Villefranche-sur-Saône were being integrated, respectively, into those of Rouen, Sarrebruck (Germany), Saint-Étienne and Lyon. Urban area fusions reflect functional realities, but they are also statistical artifacts whose effect is to eliminate territories such as those in the above-cited set from the list of French urban areas and, mechanically, to increase the size of the areas that take them in.

24All statistical constructions can be criticized, and the debate on the arbitrary nature of territorial breakdowns is “recurrent and endless” (Desrosières 2008b). INSEE’s classification system remains an indispensable tool for measuring the phenomenon of peri-urbanization overall, comparing national urban systems, and studying types of urban growth over long periods. It also constitutes a common language for studying what is in essence a complex, multiform and changing phenomenon. In any case, it is a system that cannot be ignored in conducting analyses that can later be pursued at a finer scale or extended to take account of the formation of urban regions or larger territorial networks (Veltz 2005, Vanier and Roux 2008).

Peri-urban growth and “generalized urbanization” trends

27In 2011, the peri-urban population in France stood at 15.3 million (nearly onefourth of the population of European France), 12 million of whom were living in rings of large urban areas and 3.3 million in multipolarized municipalities. This population is distributed over a vast area (38% of European France for 24% of the population) characterized by extremely low average density: 74 inhabitants per square kilometer (Table 2).

28France’s peri-urban space is less remarkable for the present size of its population than for its demographic growth over the 1968–2011 period. As we have seen, that growth was not only “internal” (i.e., through densification of existing peri-urban areas) but also “external” (integration of urban units and rural municipalities into peri-urban space). In a parallel development, peri-urban municipalities caught up by sprawl from the urban unit (i.e., the conurbation: city plus continuous suburbs) were taken into that unit and cut off from the peri-urban ring. These developments explain why the peri-urban perimeter was constantly changing, independently of INSEE’s methodological changes to ZAU construction.

29To do comparisons over time, we need to use the same zoning classification (Julien 2000, Baccaïni and Sémécurbe 2009). We therefore extrapolated backwards from our own zoning, applying it to each census wave since 1968. This required us to identify as peri-urban a number of municipalities that were not yet so according to earlier censuses. Analysis with zoning classifications kept constant therefore does not inform us about past peri-urban space but rather about demographic change in spaces now defined as peri-urban. This way of proceeding makes it possible to apprehend the demographic scope of peri-urbanization as a process, rather than degrees of peri-urbanization at a given moment.

30Figure 1, constructed on this basis, shows that demographic growth in municipalities now classified as peri-urban was strong and continuous from 1968 (+62%; + nearly 6 million inhabitants) and accelerated between 1975 and 1982. This increase in the population of peripheral space stands in stark contrast to the dynamic in cities over the same period: their population declined from 1975 to 1990, a trend that seemed to suggest at the time that a “counter-urbanization” process was at work (Berry 1976). However, in the 1990s, a considerable number of centers began to grow again (Le Jeannic 1997). Suburban municipalities, meanwhile, showed steady growth from 1968 to 2011 (+48%) though the curve gradually flattened. While population growth rates were stronger in peri-urban areas than in continuous suburban municipalities, the opposite is true of the numbers: suburban municipalities gained 7.3 million inhabitants over the period as against 6 million in peri-urban areas.

Figure 1

Changes in population by zoning type from 1968 to 2011 (millions)

Changes in population by zoning type from 1968 to 2011 (millions)

31Though the population of urban units diminished relative to that of peri-urban areas, it is still the majority: 64% of the population of European France lives in urban units now, as against 24% in peri-urban areas. In 1968, with zoning classification kept constant, the respective figures were 65% and 19%. Densification of continuous suburbs is of course no longer the only driver of urban growth, but the effect of population redistribution on urban units has been limited compared to their initial weight.

32By breaking down demographic change over the 1968–2011 period into natural increase and net migration balance (Table 3) we can better understand the forces at work. As might have been expected, peri-urban growth is due to residential mobility. Natural increase in peri-urban areas (calculated relative to the population) is lower than in centers and much lower than in continuous suburbs, but it sharply distinguishes peri-urban areas from rural ones, where natural increase is negative. By contrast, population growth in suburban municipalities is driven by strong net natural increase combined with a positive migration balance. Centers showed a negative migration balance over the period, though it was more than compensated by net natural increase.

33Clearly, peri-urban growth is driven by strong centrifugal forces. However, as suburbs lost households to peri-urban areas they were also receiving new ones. And peri-urban settlement was not and is not always centrifugal: many of the new households came from areas outside the urban areas in question. This applies in particular to southern France: Marseille-Aix-en-Provence, Montpellier and Toulouse (Le Jeannic 1997, Jaillet 2013). It also holds for the Paris urban area (Berger 2004), though not for the majority of new households there. Last and most importantly, statistical analysis with zoning classifications kept constant shows that the size of the peri-urban population should not be exaggerated. While 15.3 million people were living in peri-urban space in 2011, the municipalities that make up that space today already had a total population of 9.4 million in 1968. In other words, the peri-urban areas that received new households were hardly empty of residents. Contrary to a widely accepted assumption, we cannot explain the current weight of the peri-urban population (24% of the French population) exclusively by a “tidal wave of single-family home construction”; we also have to recognize that peri-urbanization involves integrating rural and urban areas into urban unit zones of influence. The stronger functional tie between peripheral municipalities and employment catchment areas is therefore due to both the commuting distance of people working in those units and the attraction those same units exert on working people living in municipalities undergoing peri-urbanization. The arrival of new inhabitants has invigorated those areas, making it easier for people already settled there to stay put. This explains the creation of new facilities, services and jobs at the local level. On this point we agree with Olivier Piron (2014: 31) that it is mistaken to “equate peri-urban settlement with people who have come to live there, and forget those who already live there.” This is a frequent error.

34Population numbers are not the only phenomenon affected by peri-urban development: economic activities and jobs have also been moving centrifugally. First among them are activities that consume a great deal of space or require rapid access to highways and freeways, such as logistics, the building industry and public works, mass consumption, and leisure activity sites. Most enterprise zones built on city peripheries include traditional or technological industrial activities (Wiel 2010). This phenomenon is particularly pronounced in the Île-de-France region (Gilli 2005).

35The centrifugal movement of populations and jobs facilitates the formation of new polarities within urban areas, polarities that compound those in urban units encompassed in peri-urban development. Secondary centers, either longstanding or emergent, play an important role in the residential choices of households investing in peri-urban areas and in the organization of daily lifestyles and mobility there (Kaufmann 2014, Marchal and Stébé 2015). Enterprise zones especially show a high concentration of facilities and services, and they are places of diverse population mix. Sociability in peri-urban areas is organized at different spatial scales than a housing tract or the municipality as a whole. From this perspective, peripheries seem less dominated by and dependent on the main center than suburbs (Jaillet 2013).

36The question remains whether it makes sense in these cases to speak of polycentric organization, as do promoters of “emerging cities” in France (Dubois-Taine and Chalas 1997). As Denise Pumain (2004) has pointed out, the spatial spread of business activities primarily benefits continuous suburbs, and this works in favor of peri-urbanization by ricochet, enabling people to settle still further out from cities (Le Jeannic 1997, Berger 2004). In strictly quantitative terms, European France’s 230 large urban units account for 71% of employment for approximately 60% of the population (Brutel and Levy 2011). The spatial development of employment is therefore more limited than population spread. And qualitatively speaking, there is no “spin off” of metropolitan functions, i.e., high value-added activities and jobs. “Emerging” polarities (some located in suburban municipalities) generate only partial and fragmentary functions compared to what urban centers provide. They organize the peri-urban milieu but cannot compete with urban centers, in contrast to American “edge cities.” [6] It is of course true that peri-urban space is not as single-function as is often thought. Recent INSEE analyses of “bassin de vie” [7] areas do conclude, however, that peri-urban areas primarily offer neighborhood and “intermediate” facilities and that there are few “high” level facilities such as national employment offices, high schools, emergency hospital services, maternity clinics, medical specialists, hypermarkets or movie theaters, all of which are concentrated in urban units (Brutel and Levy 2012).

Peri-urbanization intensity varies by metropolitan area

37To pursue the analysis, we now take a comparative look at the peri-urban spaces situated around France’s “metropolitan areas”; that is, the twelve urban areas at the top of the list (see Box 1). We did not take multipolarized spaces into account since by definition they cannot be attached to or identified with a single urban area. Our aim here is two-fold: 1) to show that peri-urbanization intensity varies by metropolitan area, and 2) to investigate the claim made by several authors of a tie between periurbanization and metropolitanization (Ascher 1995; Roux and Vanier 2008). [8]

38On average, the peri-urban rings of metropolitan areas represent 19% of those areas’ populations. This figure conceals marked differences between areas, however (Figure 2). The 60,000 inhabitants of the peri-urban ring of Nice represent only 6% of the area’s population; the peri-urban rings of Marseille-Aix (9%) and Lille (12%) also have relatively low populations. Rennes (54%), Strasbourg (41%) and Nantes (32%) are clearly the most intensely peri-urbanized metropolitan areas. Meanwhile, 1.8 million inhabitants of the greater Paris metropolitan area are peri-urban, but they account for only 15% of the greater metropolitan area population. In fact, the relative weight of peri-urban populations is not related to the size of the urban area or its degree of metropolitanization, an observation confirmed by comparing them with “non-metropolitan” areas, where 25% of the population, on average, live in peri-urban rings. Peri-urbanization is therefore more a process of “generalized urbanization” (Lefebvre 1970) than a dynamic particular to metropolitanization.

Figure 2

The peri-urban ring populations of France’s 12 metropolitan areas in 2011

The peri-urban ring populations of France’s 12 metropolitan areas in 2011

Note: Urban areas listed by decreasing population size

39A comparison of peri-urban ring population growth rates for France’s twelve major cities seems to suggest that a catch-up phenomenon is under way in urban areas where peri-urbanization has been relatively low. The rates vary from 51% (Strasbourg) to 215% (Marseille-Aix) (Table 4). But it would be mistaken to conclude that high peri-urban population growth rates in Marseille-Aix, Montpellier and Nice mean that those urban areas are undergoing massive peri-urbanization: rates are highest where initial peri-urban populations were low.

40The peri-urban rings of France’s twelve major cities show average growth of 52% from 1968 to 2011—nearly twice the figure for their continuous suburbs. In absolute figures, however, continuous suburb population growth was larger. Peri-urban ring population growth accounted for 35% of metropolitan area population growth while urban unit growth accounted for 65% (Table 5) Furthermore, peri-urban population growth accounted for less metropolitan than non-metropolitan area growth (46%), exceeding the 50% mark only in Rennes (65%) and Strasbourg (52%). At the other end of the scale, peri-urban growth accounted for only 11% of the growth of the Nice metropolitan area. Paris, meanwhile, came in slightly under the national average with a 31% contribution from peri-urban growth. If we limit the focus to the most recent years (2006–2011), however, we see that the contribution of peri-urban growth to metropolitan area population growth rises to 38%. This figure also indicates that the peri-urban growth dynamic is not as strong as for urban units. But once again, contrasts by area are striking. While the contribution of peri-urban growth has recently reached record levels in Strasbourg and Rennes (over 80%), it was a mere 20% in 2011 in Paris, an area marked instead by the dynamic of its suburban municipalities. Paris is decidedly an isolated case.

41Demographically, then, we see a strong differentiation between metropolitan areas— for which there can be no simple explanation. To pursue the analysis we would need to take into account the specific geography and history of each of the cities in question (Bessy-Pietri 2000). In the case of major conurbations and multipolarized systems such as Marseille-Aix, Lille, and Nice, INSEE’s urban area classification does not facilitate precise analysis of urbanization dynamics (Cusin 2016). It is more relevant for areas whose organization follows the classic center-periphery model, such as Paris and most metropolitan areas of western France (Julien 2000).

The social makeup of the peri-urban spaces of France’s twelve metropolitan areas

43We used boxplots to diagram social composition (Box 2). Figure 4 presents the overall characteristics of the twelve metropolitan areas by zoning type for a series of nine socio-demographic variables. At this scale, the areas are differentiated from each other mostly by age composition. Peri-urban areas seem to be family places: there are considerable shares of persons aged 30-59 and under 14; the respective medians are 43% and 21%, as against 36% and 15% in centers and 41% and 20% in suburban municipalities. Conversely, the share of persons aged 15–29 living in periurban areas is far lower than those in centers and suburban municipalities: a median of 15%, as against 30% and 17% respectively.

Box 2.—Boxplots: a tool for visualizing peri-urban diversity

Boxplots present the shape of statistical distributions. Each box is located between the 1st and 3rd quartile of the distribution and accounts for 50% of individuals. The horizontal line in the middle of the box represents the median (2nd quartile). The vertical lines represent municipalities with values higher than the 3rd quartile or lower than the 1st. In most cases these outlier points are located above or below the lines and correspond to observations situated at some distance from the rest of the distribution. The figure below represents the share of 15–29-year-old individuals in the populations of France’s metropolitan areas in 2011 by zoning type (Figure 3).

Figure 3

15–29-year-olds in metropolitan areas in 2011, boxplot (% of reference population)

15–29-year-olds in metropolitan areas in 2011, boxplot (% of reference population)

The box on the left shows that centers are occupied by a younger population than suburban municipalities or peri-urban areas. The median proportion of young people in centers is 30% whereas for suburbs it is approximately 18% and for peri-urban municipalities, 15%. Two municipalities do not follow this rule, however, showing particularly low shares of young people; they are the outliers. It is clear that while the share of young people varies greatly by peri-urban municipality (from below 10% to a maximum of almost 40%), the 1st and 3rd quartiles are relatively low compared to those same quartiles for centers and suburbs.

Figure 4

General profile of the population of metropolitan areas in 2011 (percentages of reference population)

General profile of the population of metropolitan areas in 2011 (percentages of reference population)

44Meanwhile, whereas office workers and mid-level occupations are homogeneously distributed across the three zoning types, managers-professionals and manual workers distinguish peri-urban spaces from other types. Peri-urban areas show sharply lower shares of managers and professionals than centers and suburbs: more than half of peri-urban municipalities have under 9% managers (median value) while nearly all centers have a higher share. However, the share of managers in peri-urban areas is higher than the average for France as a whole, and some peri-urban municipalities have high levels of managers: the maximum value is 34%—above the highest values found for centers (27%) and suburban municipalities (32%). It is therefore inaccurate to claim that the proportion of managers in peri-urban areas “belonging to” French metropolitan areas is low.

45Manual workers and managers-professionals are diametrically opposed in terms of where they live and work, confirming findings of earlier studies (Tabard 1993, Préteceille 2006). In half of peri-urban municipalities we find over 14% manual workers. The corresponding median values for cities and suburbs are lower (10% and 9% respectively). However, similarly to managers, the median share of manual workers in peri-urban municipalities is close to the national average of 14%. Variations by municipality range from nearly 0% to 40%, attesting to the overall diversity of these spaces and to local social polarization phenomena.

46If we compare shares of managers and manual workers at the municipal level, we find strongly heterogeneous peri-urban spaces. However, the considerable presence and low dispersion of mid-level occupations (also office workers) suggests that the majority of spaces are “middle-mixed,” to quote Edmond Préteceille (2006).

47To complete this socio-occupational overview (using INSEE’s Professions et Catégories Socioprofessionnelles classification system), we now look at living standard measurements. Recent figures published by INSEE show that peri-urban rings and multipolarized spaces have higher living standards than major metropolitan area centers and suburbs (Aerts, Chirazi and Cros 2015). Their poverty rate (8.8%) is much lower than in centers (19.5%) or suburbs (13.9%). It therefore cannot be claimed as Guilluy does (2010) that peri-urban areas as such are spaces of social banishment in which only the economically insecure or impoverished live. If peri-urbanites are on average financially more comfortable than inhabitants of urban units—albeit with marked differences by distance from the center—this is in part because the proportion of working people in them, especially working couples, is higher (Roux and Vanier 2008). Nor can we say that peri-urban space is occupied primarily by the middle-class. As shown by living standard measurements, this living area is actually a “middling” one compared to centers and suburbs, in that the entire range of socio-occupational categories is represented (though proportions vary by municipality) and there are lower shares of households from the two extremes of the social hierarchy (Berger 2004, Jaillet 2013). This is confirmed by the fact that social inequality is lower in peri-urban areas than in the other two zoning types (Aerts, Chirazi and Cros 2015).

The coherence and heterogeneity of peri-urban development

48We now pursue this socio-demographic approach by way of principal component analysis (PCA), which allows for summarizing complex relations between a high number of variables. As a “tool for exploring and visualizing the internal variability of a population characterized by many variables” (Desrosières 2008a), PCA places individuals on factorial “axes” in order to identify which features structure social phenomena. Each axis synthesizes a segment of relations between variables included in the analysis and measures the distance between individuals. PCA is particularly well suited to the study of phenomena whose structuring dimensions are not very well known (Amossé and Chardon 2006), enabling researchers to test whether or not a statistical category accurately groups together individuals and differentiates them from other categories.

49Our PCA differentiates each of the metropolitan areas from the others and treats non-metropolitan areas and spaces outside urban areas as a single entity. For each urban area we distinguished between the center, its continuous suburbs and its periurban ring. To examine population distribution we chose the active variables of age, socio-occupational category, educational attainment, household type, family size and employment characteristics. The first two axes account for 83% of total inertia; that is, of the entire set of relations between the spaces chosen for analysis as determined by these variables. They bring to light a dual structuring opposition (Table 6).

50Axis 1 establishes an opposition between spaces inhabited by the young and highly skilled (aged 15–29, students, noncohabiting persons, senior managers and professionals, highly educated individuals) and those inhabited by the working class (manual workers, vocational certificate holders) and the family-oriented (aged 40– 59, couples with children). Axis 2 establishes an opposition between spaces inhabited by senior managers, highly educated people and/or couples with children, and spaces with exclusively working-class and poorer inhabitants in which large families and individuals with little or no education, unemployed or without an occupation are overrepresented (Figure 5). Combining the two axes generates a “complete system of relative positions” (Desrosières 2014) of the different spaces comprising European France.

Figure 5

Factorial representation of the different spaces of European France

Factorial representation of the different spaces of European France

51We see that Axis 1 establishes an opposition between centers, on the right, and peri-urban rings, on the left, with suburbs in the middle. The center-periphery opposition is more significant on this axis than whether or not the zone belongs to the same urban area—a finding that confirms the socio-demographic specificity of peri-urban space. Axis 2 distinguishes peri-urban rings: it separates peri-urban spaces from metropolitan areas, located in the lower half of the figure, and from non-metropolitan spaces, located in the upper half. The intersection of the two axes circumscribes the peri-urban rings of metropolitan areas in the lower half of the figure: they clearly form a distinct category. The peri-urban spaces of metropolitan areas are therefore more similar to each other than to centers and suburbs: they differ not in that each belongs to a particular urban area but rather in that the socio-demographic profile of their inhabitants contrasts directly with that of inhabitants of centers and suburbs. But the peri-urban as a spatial-social category cannot be reduced to a homogeneous peripheral space. The position of peri-urban rings on Axis 2 shows that they are also spaces with diverse socio-demographic profiles whose specificity lies not so much in the decidedly real weight of the working-class and family-oriented population than in the co-existence of working-class inhabitants and highly educated working people, senior managers and people in mid-level occupations.

52The PCA reveals the coherence of the peri-urban as a space that differs socially from both centers and suburbs. But it also brings out internal differentiation factors. Axis 1 shows that the peri-urban rings of Nantes and Bordeaux have relatively peripheral socio-demographic profiles while those of Montpellier, Marseille and Lille are very similar to the profile for metropolitan area suburbs. Axis 2 establishes an opposition between the peri-urban rings of Lille and Grenoble, which are extremely distant from non-metropolitan spaces, and those of Nice and Bordeaux, located at the same level as the peri-urban spaces of non-metropolitan areas. Meanwhile the peri-urban ring of Paris does not differ significantly from those of other metropolitan areas. However, the Paris metropolitan area is sharply distinguished from the others by the extreme coordinates of the center on both axes. The position of the Marseille-Aix main center (i.e., Marseille) contrasts with that of Paris, but their respective peri-urban rings are very close. Generally speaking, metropolitan and non-metropolitan urban areas differ less in terms of the socio-demographic characteristics of their peri-urbans ring than by the socio-demographic characteristics of their center cities.

53This approach also brings to light internal differences between metropolitan areas by peri-urban ring. We introduced each peri-urban municipality of the metropolitan areas as an additional individual into the correspondence analyses, thereby attributing coordinates to each on the axes identified by the analysis. We then calculated the standard deviation of the coordinates for the municipalities of each metropolitan area peri-urban area to measure municipality dispersion within them (Table 7).

54On both Axis 1 and Axis 2, Grenoble’s peri-urban municipalities are the most heterogeneous, while Nantes’ are the most homogeneous. Peri-urban Grenoble is composed of municipalities whose inhabitants have center-identified (right side of Axis 1) and metropolitan (lower half of Axis 2) socio-demographic characteristics and municipalities with the opposite profile. The values for inter-municipality dispersion within peri-urban rings sheds new light on the socio-demographic mix particular to these spaces. We find a certain degree of social mix in the peri-urban rings of Nantes and Lille, whose municipalities are not widely dispersed on the two axes. The results for Grenoble and Toulouse, however, can be interpreted as a kind of socio-spatial segregation that gets played out at the municipal scale through the coexistence of municipalities with strongly center-identified profiles and diversified inhabitants, and municipalities with more periphery-identified profiles and a more segmented population. The PCA thus reveals the coherence of “metropolized” periurban areas compared to other spaces. Nevertheless, the internal diversity of periurban space should preclude our thinking of it as a homogeneous block.

Polarization of managers-professionals and manual workers varies by urban model

55To identify in greater detail the socio-demographic differentiations between peri-urban spaces and bring to light finer distinctions between metropolitan areas we developed a “boxplot” representation of proportions of managers-professionals and manual workers (the two most discriminating socio-occupational categories) at the municipal scale for each of the metropolitan areas in question. [9]

56Paris, Lyon, Bordeaux and Nantes show a decline in the median proportion of managers-professionals as we move away from the center (Figure 6).

Figure 6

Proportion of managers-professionals in metropolitan area municipalities by zone type in 2011 (percentages of reference population)

Proportion of managers-professionals in metropolitan area municipalities by zone type in 2011 (percentages of reference population)

Note: Reference population over age 15.

57This configuration corresponds to the housing market price structure of the “classic” center-periphery model: highest prices in the center, lowest in peri-urban areas (Cusin 2016). Lille, Montpellier and Nice, on the other hand, do not follow that model: the median proportions of managers-professionals in their peri-urban rings are quite close to those in the center. Meanwhile the median proportion of senior managers in the Marseille-Aix area is higher in “peri-urbia” than in the center city.

58As we have seen, for all metropolitan areas taken together, the polarization of manual workers follows the opposite pattern to that of managers: median proportions are higher in peri-urban areas than in other zone types. This is not an absolute rule, however. In fact, the proportion of manual workers in peri-urbia varies considerably by metropolitan area, ranging from 9% for Marseille to 18% for Nantes (Figure 7). Just as we found for managers, some areas diverge from the classic center-periphery model, particularly in the metropolitan areas of Marseille, Lille and Montpellier, where shares of manual workers in peri-urban rings are lower. In Marseille, for example, the median proportion of workers in peri-urban municipalities is 9%, as against 10% in the center. And Marseille reverses the classic center-periphery model with regard to both socio-occupational groups, in that social selectivity and housing prices tend to increase with distance from the center (Cusin 2016). In the Lille metropolitan area, for example, the median proportion of manual workers living in peri-urban municipalities (10%) is lower than in suburban municipalities (11%), while in Montpellier and Nice that proportion is fairly close to the one found for the center. Generally speaking, peri-urbia in France is not characterized by social homogeneity, as attested by the proportions of mid-level occupations and office workers. We should also be cautious about claiming that peri-urban space is strongly working-class given the magnitude of disparities between municipalities. In the twelve peri-urban rings, only a handful of municipalities have over 30% manual workers, and the figure is below 20% for three-fourths of municipalities.

Figure 7

Proportion of manual workers living in metropolitan area municipalities by zone type in 2011 (percentages of reference population)

Proportion of manual workers living in metropolitan area municipalities by zone type in 2011 (percentages of reference population)

Note: Reference population over age 15.

59Mapping proportions of managers-professionals and manual workers brings to light the social mosaic of French peri-urban space. To illustrate this, we have chosen two metropolitan areas with strongly different profiles: Lyon, which follows the classic model, and Marseille-Aix, which offers a directly contrasting model.

60Maps 3 and 4 of Lyon (Appendix) show the polarization of managers between the center and the outermost municipalities (especially western ones), while municipalities with considerable shares of manual workers are clearly located in peri-urban areas (mostly in the east). Nonetheless, there is not a single peri-urban municipality with over 27% manual workers, and for most of those municipalities the figure is below 19%. If we add that a number of peri-urban municipalities have a relatively high proportion of managers-professionals, the image of a social mosaic comes to the fore—and this despite the resonance of a real estate structure where prices fall with distance from the center (Cusin 2016).

61Maps 5 and 6 of Marseille-Aix (Appendix) show how narrow the peri-urban ring is, as is its demographic weight in the urban area. Similarly to Lyon, however, a social mosaic predominates. Proportions of managers-professionals and manual workers are similar from one municipality to another. The general context shows strong social polarization within the conurbation: manual workers in the west, near the Étang de Berre, and managers-professionals in Aix-en-Provence and the outermost municipalities. In contrast to Lyon, the main center of this metropolitan area, i.e., the City of Marseille, shows average proportions of both managers-professionals and manual workers. This structure, specific to Marseille, is due to the polycentric organization of this particular metropolitan area and the relatively low value of the center compared to the periphery.

62* * *

63Studies of peri-urban areas in France often take an overly general perspective— the peri-urban treated as a single bloc—or an overly particular one, as illustrated by the many monographs on the subject, whose qualitative findings are rich but fragmentary. Here we have made use of INSEE’s zoning classifications—after discussing their contributions and limitations—in the aim of renewing scientific debate and revealing the diversity of peri-urban worlds in France.

64We have shown that while peri-urbanization is indeed a major phenomenon, it dilates the city rather than dilutes it. Despite a very real “urban exodus,” the idea that peri-urbanization is occurring at a massive scale must be qualified. Urban unit populations are growing and there is no threat to their economic centrality. Meanwhile, peri-urban development and life can hardly be reduced to single-family residential areas and widely spaced housing; they orbit around urban units, and around urban units functioning as secondary centers.

65Peri-urban diversity is not just morphological; it is also social. While the considerable shares of manual and office workers living in peri-urban rings goes against the image of this space as the chosen habitat of the middle class, peri-urban space is not a place of “banishment” either. The share of mid-level occupations is approximately the same as in centers and suburbs, and managers, though underrepresented in peri-urbia, are not absent from it. Moreover, peri-urban areas have been found to have higher living standards than centers or suburbs (Aerts, Chirazi and Cros 2015). Poverty rates and inequality levels are lower there. Guilluy’s idea (2010) of a social divide between peri-urban areas and metropolitan area urban units has therefore not been verified.

66The peri-urban cannot be identified with any particular socio-occupational category in France. As in the case of urban units, it is composed primarily of “middle-mixed” spaces (Préteceille 2006) from which marked types of socio-spatial segregation, gentrification and social distress are absent—in contrast to centers and suburbs. Peri-urban areas in France overall do have a working-class “tonality,” but they also include a larger proportion of homeowners and of working people integrated into urban unit economies. The predominance of “middle-mixed” municipalities does not mean there is no social division of space—such division may be quite pronounced at a micro-local level, as we saw with municipalities that polarize managers-professionals and others that polarize manual workers. But rather than conclude from this that peri-urban areas are places where like mixes with like, we can see it as the result of generalized urbanization: urban units do not just project populations into peripheral areas; they also project socially-driven residential grouping patterns. These phenomena are particularly strong because they impact housing prices at micro-local levels and may be used by local public authorities to develop incentives. However, they are not exclusive to peri-urban development.

67Last, if we compare larger urban areas we find different differentiation factors. Peri-urban sections of metropolitan areas differ from peri-urban sections of other urban areas by socio-demographic characteristics: the shares of managers-professionals, highly educated and working people are higher in them. However, that difference is much less sharp than the one found between peri-urban areas and other urban area spaces. Factorial analysis shows that peri-urban rings have characteristics in common with each other that differentiate them from centers and suburbs—to such as degree that the “periphery effect” seems stronger than the effect of belonging to one or another urban area. Comparing metropolitan with other urban areas shows that it is not so much the social composition of metropolitan areas’ peri-urban rings that distinguishes the two groups as the social composition of their centers. As a statistical category, then, peri-urban space exhibits strong internal coherence while working effectively to discriminate not only in spatial but also social terms. It is this ability to bring out general characteristics without effacing particular cases (Desrosières 2008a) that confirms a posteriori the heuristic value of INSEE’s zoning classification.

68We have also shown through statistical analysis that peri-urbanization intensity does not depend on degree of metropolitanization (cf. the comparison of metropolitan and non-metropolitan areas). Peri-urbanization thus seems a more general phenomenon than metropolitanization. Ultimately, the peri-urban rings of major metropolitan areas differ from each other. These differences are explained by the histories of the particular cities, histories through which city development was shaped by the geographical properties of the land, local economic systems and changes in them, residential models, and urban policies. In the end, it is not so much the statistical category of the peri-urban that should be called into question as the theoretical frameworks used to account for the reality of peri-urban development. Far from analyses that sum up peri-urbanization in France as a phenomenon of socio-spatial banishment or middle-class secession from cities, it would seem that peri-urbanization goes not against the city but in close relationship with it, all the while contributing to changes in city scale and socio-spatial composition.