1. Introduction and problematic

2Moreover, most analyses tend to privilege observation of economic concentrations, giving secondary attention to demographic concentrations [Leroy, 2000]. Consequently, on the one hand, the circulation of people is chiefly analysed with regard to human resources envisaged as contributing added value to the process of concentrating wealth and power, a process which is at the heart of metropolisation [Veltz, 1996, Da Cunha and Both, 2004]. The focus is then on qualified and degree-holding persons of the tertiary sector (even if certain studies compare different social categories – see, for example, Huber 2014), on the model of public migration policies, which, to do this, speak progressively of international mobility [Pellerin, 2011], then finish by distinguishing this international mobility from international migration, in order to gather around this new term migrations that are legal, planned, safe, and economically profitable [Borja and Ramadier, 2014]. On the other hand, analyses (for example Donzelot, 2004, 2009, Maurin, 2004, Fusco and Scarella, 2011) of the social segregations produced by metropolisation have appeared in a second phase, after the processes of urban attraction/sprawl, of concentration and selection of activities of governance in fields as varied as politics, economics and culture have been the object of thorough investigations. Now, how do the socio-residential transformations of the circumferences of a metropole function when it is separated from its area of geographic influence by a national border?

3Our point of departure is the following idea: the socio-residential evolution of persons located outside the metropole of Luxembourg, but who are nevertheless professionally fixed in the metropole (the metropolitan area of influence), can help us to shed light on the socio-spatial issues associated with metropolisation. In the case of Luxembourg, the frontier with France reveals possible social inequalities or indeed disparities in the socio-demographic evolution of the different working populations (resident and cross-border). Thus this is also a veritable methodological tool, permitting us easily to identify the majority of the metropolitan employees located outside the metropolis, and, on the other hand, to analyse the particular case of a portion of the metropolitan space whose socio-residential changes are amplified by that geopolitical boundary which separates, even as it favours people’s contact with the centre of the metropole [Lord and Gerber, 2012].

4Yet it is not because the political frontiers have evolved and permit other economic and cultural practices previously difficult to envisage that relations to the geographic space are completely transformed. If these evolutions have permitted the emergence of veritable cross-border metropoles, like that of Luxembourg [Sohn, 2012], studies of the attachment to place of the Luxembourg frontier commuters show that certain former cultural anchorages persist in current cross-border policies [Enaux and Gerber, 2008]. By contrast, studies of residential mobility confirm that this frontier reinforces the socio-economic disparities of metropolisation by encouraging the most well-off employees to choose residentialisation in the Luxembourg capital [Lord and Gerber, 2012, Diop, 2011], at least with regard to recent decades. Consequently, in what measure do the socio-spatial disparities of a metropolitan region depend both on previous socio-spatial structures and on the current socio-economic processes of the territory? Likewise, in what measure can metropolisation not be simply reduced to the attraction of ‘acteurs de l’innovation’ (promoters of innovation)? In other words, how are geo-residential disparities organised between the cross-border lower and ‘middle’ classes?

5This investigation of the evolution of socio-residential distributions of cross-border commuters from the French part of the metropolitan region of Luxembourg supplements the recent studies conducted by the French national statistical agency (INSEE) [Floch, 2015] and pre-supposes that:

• the post-industrial city affected by metropolisation is not limited to the ‘classes créatives’ (creative classes) (Florida, 2005), executives, managers and their headquarters. Certainly, a number of studies show that metropolisation especially attracts employees who hold degrees and are oriented towards supervision and management. But are the working classes therefore absent from the social equation? Does spatially relegating the working classes [Donzelot, 2009] mean their exclusion from the stakes of metropolisation? The general assumption is that these social groups contribute no ‘added value’. Thus, on the one hand, commentaries and analytic essays, notably by the ‘Los Angeles school’ [Dear and Flusty, 2002], tend to substitute cultural pluralism (lifestyles, cultural origins and references, etc.) for sociological structures. On the other hand, social relations are no longer analysed, notably because they are considered to be less readable than formerly [Hamel, 2010]. Now, we consider that analysis of social relations on the basis of the geographic (and especially residential) space enables us to supply elements of understanding of the processes of metropolisation which are currently neglected. This leads us to propose, first of all, the hypothesis (1) that, if the cross-border working classes of the metropolitan region of Luxembourg, in particular the frontier workers who initially participate significantly in the Luxembourg economy, then in the metropolisation, diminish proportionately over the course of recent years, they nevertheless continue to increase in raw numbers. Consequently, as a result of metropolitan growth, the cross-border workers diminish proportionately less than non-cross-border workers in the area of metropolitan influence. Thus, just as ‘il n’y a pas un mais des périurbains’ (there is not one but a number of peri-urban areas) [Charmes et al., 2013], there is not one but a number of cross-border areas of metropolitan influence.

• metropolisation has not increased residential mobility as much as is sometimes stated. Certainly, commuting journeys are constantly longer in time or in kilometres, and the automobile is the favoured means of travel – especially, moreover, for cross-border workers [Schmitz et al. 2012] – but residential mobility outside the metropolis conforms to forms of social logic [Debroux, 2013]. These spaces are not the playground of middle classes who settle for an in-between both sociological (neither just getting by nor well-off) and geographic (neither urban, nor suburban, nor rural), as proposed by Marie-Christine Jaillet (2004). The individualised principle of elective affinities (choice of residence, for example), whether in its social or spatial dimension, would thus encounter limits, because residential location would be in part the result of social relations taking shape in the geographic space. The underlying hypothesis (2a), then, is the following: with the proportional increase in qualified jobs, the residential spaces which were formerly destined for cross-border employees from the working-classes would now receive greater proportions of more highly qualified people. In other words, the growing value placed on highly qualified cross-border jobs would translate socio-spatially in terms of ‘residential choice’. Thus, the sociological evolution of cross-border workers would partially correspond to their residential evolution. Only partially, because the demographic growth of cross-border employees over the decades is necessarily accompanied by a growth in the urban area. Thus, as the urban metropolises have progressively integrated the surrounding village centres, the metropolitan regions progressively integrate the secondary urban centres (namely Thionville, Longwy, even Metz in the case of the Luxembourg metropolis). Now, these secondary urban centres represent an opportunity of living in the city without living in the metropolis. On this basis, we also propose the hypothesis (2b) that if the metropolisation of Luxembourg is a function of economic activity, the secondary urban centres of this metropolitan areas contribute to the capital of the Grand Duchy by taking part in a metropolisation heavily dependent on the dwelling-places of the most qualified frontier workers. Thus, the secondary urban centres, like the metropolises, would experience socio-economic growth focused on the gentrification of their population and an ‘économie résidentielle’ (residential economy) [Davezies, 2003 ; 2008] which, at the level of a metropolitan area, joins the productive economy to the residential economy [Davezies and Talandier, 2014].

• metropolisation is not a process more strongly conducive to ‘socio-spatial fragmentation’ than other processes of urbanisation or residentialisation. On the one hand, socio-spatial segregations already existed in that portion of geographic space which interests us, even if it had another socio-economic status (rural setting, small city, etc.). Thus, contrary to the widespread idea that metropolisation upset our representations of space and our relations to it [Vanier, 2013], because it supposedly became the standard everyone applies in renewing the meanings of geographic space [Bassand et al., 2001], we adopt instead the idea that these are mainly socio-residential reconfigurations occasioned by the sociological reconfigurations of border-crossing employees. Hence, there would be neither, on the one hand, a geographic space void of all social (and functional) disparity, which found itself ‘torn apart’, fragmented by local socio-economic and socio-political activity, nor, on the other hand, segregations fluctuating according to the renewal of spatial meanings and practices, with no connection to the previous socio-spatial structures. Consequently, we posit the hypothesis (3) that the processes of residential segregation of metropolisation are interactively engaged with the processes of segregation active previously. Thus, the social history of the secondary urban centres models their process of gentrification as previously described. The pace of gentrification differs from one secondary urban centre to another (for example, Longwy would attract highly qualified cross-border employees less rapidly than Thionville, because of its greater working-class population in the past) – a fact which contributes to maintaining the socio-economic hierarchies of the secondary cities in the metropolitan area.

6In order to respond to these three main hypotheses, we well attempt, in a second section, to contextualise the study perimeter of the Luxembourg-France frontier, especially in light of the data available in French censuses. This will also make it possible to establish a perspective on the structural evolution of the working population concerned from the end of the 1960s to the present, while setting out the methodology employed. It will then be possible to test our hypotheses in a third section, which will focus more on analysis of the socio-residential evolution of the cross-border employees going to Luxembourg. The final concluding section will be devoted to discussion of these results, as well as to indicating certain directions for research.

2. Framing data and methodology

2.1 Evolution of workers in Lorraine since the 1960s

8Very few studies, however, have focused on a comparison between this frontier population crossing into Luxembourg and local employment in Lorraine, except, perhaps, for a series of papers in Economie Lorraine derived from the INSEE (the most recent being that of François and Moreau, 2010). Now, the figures speak for themselves: it must be acknowledged that the active population of the three départements in our space of reference working in France still form by far the overwhelming majority in 2013 (Figure 1). Whether at the beginning or the end of the period covered by the census, they represent 99% of the working population in 1968, although diminishing to 88.2% in 2013.

9Frontier-crossing workers therefore become more and more numerous in certain départements of Lorraine, the census data of the French population making it possible to furnish an exhaustive panorama, from the 1960s to the present, of the socio-demographic and spatial evolution of the entire working population of Lorraine. Thus it is possible to distinguish three main categories of frontier-dwellers: those working in Luxembourg, others working either in Germany or Belgium, and finally, most numerous, those employed in France (Figure 1). The diminution of the proportion of workers employed in France occurs essentially in favour of the border-dwellers working in Luxembourg. The proportion of these has greatly progressed over this period, passing from 0.3% at the end of the 60s to 8.7% in 2013, with a large increase beginning in the 1990s. These statistics alone attest to the establishment of the process of Luxembourg metropolisation which took place across the frontiers at the end of the second millennium [Sohn 2012]. The other cross-border workers, those having their place of work either in Germany or in Belgium, also experience a proportional increase, but one much more moderate than those working in Luxembourg during the dates in question. Let us take the case of Germany.

10From the 1960s to the 1980s, the German job market was very attractive, offering advantageous salaries to workers and so drawing more border-dwellers from Lorraine to work there [Buxeda, 2005]. Thus between 1968 and 1982, the number crossing into Germany totalled 29,231 persons, or double the number of border-crossers going to Luxembourg. After the 1980s, when recruitment began in metropolitan Luxembourg, the number as well as the proportion of those crossing into Luxembourg began to increase sharply: for example, the raw number of French border-dwellers working in Luxembourg rose from 6,528 (0.8% of the total working population) to 14,628 (2.1%), while those working in Germany reached 15,988. Between 1990 and 2013, the growth in border-dwellers going to Luxembourg was dazzling, rising to 70,013 individuals at the end of the period; those crossing the border to Germany would not exceed 19,000 individuals in 2013.

11Finally, the last category of workers, that is, those working internationally, remained static at a marginal level, fluctuating between 0.01 and 0.2% of the proportion of workers over the period considered. This category will no longer be mentioned in the analyses that follow.

2.2 Harmonisation of the data sources and spatial categorisations

12In order to go further in its exploitation and socio-spatial analysis, the large statistical base of the French census still requires certain processing aimed at assuring some harmonisation of the socio-demographic criteria. This notably concerns files at the local level, where the level of education or the socio-professional categories must be associated, to enable them to be compared over time. For example, with regard to level of education, a simplification of the oldest census record has been necessary, even while granting that these possible consolidations undoubtedly present some limits (see annexe 1). Of 12 different levels of education in 1968, it was necessary to reduce these to 4 levels, such as figure in 2013, in order to assure the comparison, namely: i) No diploma or at most lower secondary level; ii) Lower secondary level achieved; iii) Secondary level achieved (Baccalauréat in French, all types combined); iv) Higher education degree. Of course, this process results in a loss of information with the passage of time, notably with respect to those without a diploma, but also those having a diploma superior to the baccalauréat (the nomenclature not taking account of the first, second or third cycles). This example of simplification reflects, moreover, that applied to the socio-professional categories, where we pass in 1968 from 18 types to the 4 finally adopted in 2013 in the framework of our study: i) Managers and higher intellectual professions; ii) Intermediate professions; iii) Employees; iv) Labourers.

13Nevertheless, these data, transformed and certainly reduced in value, still make it possible clearly to establish a picture of the evolution of the socio-demographic of the residents, both border-crossing and not, of the three départements of our space of reference, as we will see in the following section.

14Finally, in order to facilitate the socio-spatial analysis needed for the verification of the hypotheses, it is also necessary to describe the implementation of the spatial categorisation of our space of reference. To do so, it is important to take especially into account:

• the minimisation of the time-budget for transportation to the workplace. In order to minimise this cost, cross-border workers have shown a tendency to establish themselves closer to the national frontier [Carpentier et al., 2011], whatever their country of origin, yet without crossing it to live in Luxembourg [Lord and Gerber, 2012]. Consequently, in the case of the Lorraine borderland, the closest towns adjacent to the France-Luxembourg frontier witness a disproportionately high representation of cross-border workers;

• good connections to transportation networks. Thus certain relatively distant communities enjoying a good connection via highway or rail networks, such as Thionville and Metz, may compensate for their distant location by acquiring more border-crossers possessing a relatively large economic capital;

• distance to the closest city-centre, a factor also considered in decisions as to residence [Authier et al., 2010]. The underlying logic especially extends to the presence of amenities and the facility of social contacts, as well as the symbolic value of the place of residence, which contributes to assuring adherence to a social group (Debroux, 2011) ;

• the cost of housing in relation to the cost of transportation (Alonso, 1964). This economic balance may entail distance from the metropolis when the real-estate and other housing costs are relatively high, whether for rental or purchase.

15In keeping with these forms of logic, a regional categorisation of the space of reference is proposed, while allowing for administrative limits. Three exclusive categories have therefore been established:

1. A group of ‘Urban agglomerations’ (established based on the administrative situation in 2013). These agglomerations represent the secondary urban centres in Lorraine of the metropolis of Luxemburg: they are Thionville (avec 13 communes), Metz (44 communes) and Longwy (21 communes).

2. A group of ‘Constant hosting communes’: these are municipalities constantly inhabited by border-crossing workers in all of the seven censuses. This category includes 42 communes.

3. The last consists of ‘Current hosting communes’. This group includes all the rest of the communes inhabited by the border-crossing workers; these are a fluctuating number of communes varying according to the census year.

16Figure 2 illustrates the distribution of these different spatial categories of the Lorraine borderland, all years taken together. This spatial typology will serve, among other purposes, as an analytic base for hypotheses 2 et 3. This spatial categorisation therefore corresponds to the geographic distribution of border-crossing communes. Actually, the ‘constant hosting communes’, which depend essentially on the number of border-crossing residents, include the communes which participated as much formerly as they do today in the dynamic of the metropolisation, such as Villerupt, Audun-le-Tiche and Cattenom, close to the France-Luxembourg frontier. The ‘current hosting communes’ are especially spread out on the spaces distant from the frontier or dispersed within the space under study.

3. Socio-spatial and temporal results

3.1. Evolution of the distribution of cross-border workers on the territory of Lorrain under study

17The literature on the subject of residential mobility with respect to Luxembourg border-dwellers has focused on those workers possessing considerable capital [Diop, 2011], while other authors have observed the residential inequality in the highly globalised urban zones [Lord et al., 2014]. Our starting point here is a narrower spatial scale, that of the commune, in order better to understand the residential mobility of cross-border workers in Luxembourg with the help of the socio-spatial categories previously defined.

18The geographic sprawl of the places of residence of French frontier-dwellers working in Luxembourg is shown first by the evolution of the number of communes which host them during the period under consideration (see annexe 2). Thus the category of ‘current hosting communes’ passes from 153 in 1968 to 308 in 1990, finally attaining 642 in 2013 (see Figure 2 to measure the extent of this sprawl). This last spatial category of current hosting, moreover, involves greater and greater distance from the French-Luxembourg border over time, as has already been noted in another study [Drevon and Gerber, 2012]. If the raw number of the communes affected already seems high, the evolution of the proportion is nevertheless more impressive. Of a total of 1,824 communes in 2013 within the Lorraine borderland, 35.2% of the communes are inhabited by Luxembourg border-crossers at the present day.

19Further, such sprawl is also evident from the demographic weight of each commune. In charting the distribution of cross-border workers according to the ellipses weighted by the portion of frontier-dwellers commuting to Luxembourg in the working population of the commune of residence (Figure 3), we observe that the evaluation of the size of the ellipses corresponds well to the extension previously mentioned. Between 1968 and 2013, a progressive slippage takes place from the centre of the ellipse towards the south of the region being considered. Thus, while the centre of gravity was situated towards Thionville at the end of the 1960s, with an ellipse being pulled by Longwy to the north, this centre of gravity has a tendency to displace itself towards the south to join the agglomeration of Metz in 2013, with a form of ellipse wider and more massive.

20The evolution of the proportion of cross-border workers in each spatial category reinforces this logic of dispersion (cf. also Table 1). This reinforcement is manifested first by the sustained growth of cross-border workers in the ‘current hosting communes’: amounting only to 8.1% of the total number of cross-border workers going to Luxembourg in 1968 (avec 172 persons), the population doubles in 1990 (16.1%, or 2,356 persons), finally reaching 32.4% (22,698 persons) in 2013. Moreover, the dispersion also translates into the distribution in proportion between the spatial categories. In 1968, the ‘constant hosting communes’ account for 74.4% of the cross-border workers; they subsequently lose the portion of frontier-dwellers who are essentially redistributed in the agglomerations. In 2013, the ‘constant hosting communes’ retain only 29.9% of the France-Luxembourg cross-border workers.

Table 1. Demographic evolution of cross-border workers according to each spatial category

	1968	1975	1982	1990	1999	2008	2013
Constant hosting communtes	74.4%	63.7%	59.9%	45.1%	34.7%	31.0%	29.9%
Agglomeration of Thionville	5.7%	6.0%	8.1%	14.1%	18.4%	16.8%	16.4%
Agglomeration of Metz	2.1%	2.9%	2.0%	2.7%	6.2%	7.2%	7.5%
Agglomeration of Longwy	9.7%	16.8%	15.3%	22.0%	16.8%	13.8%	13.8%
Current hosting communes	8.1%	10.6%	14.7%	16.1%	24.0%	31.1%	32.4%
Total (%)	100.0%	100.0%	100.0%	100.0%	100.0%	100.0%	100.0%
Total in raw volume (n)	2112	5890	6140	14628	35720	61442	70005

21The evolution of the workforce also reveals a tendency of cross-border workers to concentrate in the direction of the secondary urban centres. In a phase preceding metropolisation (the 1960s), the three agglomerations of Thionville, Metz and Longwy gathered only 17.4% (368 individuals) of the frontier workers; a peak was attained in 1999, with 41.3% (14,760 persons) before a falling back to 37.3% (26,365 persons) in 2013.

3.2. Temporal evolution of cross-border and non-cross-border Luxembourg workers

22Despite the Luxembourg metropolisation, the temporal evolution of socio-professional categories, as harmonised, between the cross-border workers working in Luxembourg and other persons employed in our study zone, signifies that the working-class border-dwellers living in France are more numerous until 1999, with 44.2% of the workforce (cf. Figure 4). Certainly, this large portion of working-class border-dwellers dwindles in the years 2000 and falls as far as 29.6% in 2013, but with the number employed increasing (from 15,768 to 20,690 between 1999 and 2013, that is, an increase of 24%). The proportional decline of working-class employees is also observed among the non-border-crossing inhabitants of Lorraine during the same period, passing from 29.1% à 23.6%; here the numbers drop by nearly 18%, passing from 196,660 to 166,946. The dynamics of working-class manpower are therefore not quite the same, depending on the centres of employment (France vs. Luxembourg). The opportunities of the job market in Luxembourg, especially following the closing of steel-mills and, to a lesser extent, of coal mining, notably in Moselle, make it possible to take in a part of this workforce in the 1980s [Auburtin 2005]. Finally, in 2013, the portion of working-class employees from Lorraine working in Luxembourg remains much greater than that of non-border-crossing workers in Lorraine, thereby proving that the working classes are not absent from the socio-spatial factors of metropolisation. These results therefore confirm the first hypothesis.

23This high proportion of working-class employees finally leaves little room for the other categories of cross-border manpower, with the managers and higher intellectual professions not crossing the threshold of 10% until the end of the 2000s, to reach almost 13% in 2008, that is, the same level for the non-border-crossing employees of the zone under study. The border-crossing managers have recently passed beyond this level to reach 15% in 2013, or two points higher than the non-border-crossers. This change shows that one of the principal processes of metropolisation, about which researchers agree – namely the growth in qualifications (here the level of diploma) in keeping with professional and economic management functions – applies to the city of Luxembourg. It may be noted that this process seems to have been maintained despite the so-called ‘financial crisis of 2008’. But this process should not obscure the contribution of other social classes, as has been shown above with working-class manpower. It is also important to note that the intermediate categories follow more or less the same course as the managers and intellectual professions. The border-crossers among the intermediate classes amounted to 18.4% in 1999, to reach a level of 23.6% in 2013, while the non-border-crossers evolved positively, to be sure, but less rapidly, passing from 23.7 to 26%, thus surpassing the proportion of working-class employees beginning in 2008.

24Nevertheless, despite the relative decline in the proportion of working-class employees during the Luxembourg metropolisation, whether border-crossing or not, a contrast appears in the demographic evolution between the non-border crossing and border-crossing workers in Luxembourg. The raw number of working-class non-border-crossers diminishes, passing from 19,660 to 16,946 persons between 1999 and 2013, whereas the number of working-class employees employed in Luxembourg does not cease to increase. This group of workers accounts for 15,768 persons in 1999 and rises to 20,690 at the present time. Thus our first hypothesis seems to be confirmed in part. On the one hand, it is true that working-class border-crossers increased in raw numbers despite their reduction as a proportion; on the other hand, it is equally true that the working-class border-crossers occupy, with more qualified employees, the largest segments of the workforce among the salaried border-crossers. By contrast, the proportion of cross-border working-class employees falls more sharply between 1975 and 2008, notably because the portion of this border-crossing workforce was initially greater than that of the non-border-crossers of the region under study.

25These socio-professional evolutions are also reflected in those concerned with the level of education (Figure 4). Let us simply take the example of those holding upper-level diplomas. Virtually absent among border-crossers until the beginning of the 1990s, their proportion came to exceed the third part of those holding diplomas in 2008, as opposed to less than 30% for the employees of the region working in France. In 2013, this gap widened further: the border-crossers holding higher diplomas were nearly 42%, while they did not exceed more than a third for employees in France. This strong increase in the level of education of the border-crossers occurred largely at the expense of those who did not hold at least the baccalauréat. But there again the proportion of the latter remained substantial, since in 2013 they represented 39.9% of the working cross-border employees (compared with 46.2% of non-border-crossers and those not holding the baccalauréat).

3.3. A socio-demographic perspective on the spatial categories

27Figure 5 illustrates the evolution of the raw numbers of these social classes of border-crossers in each spatial category established. A number of conclusions necessarily follow:

1. • 5 If one limits oneself to the analysis of 2013, the gap is greater between the number of privileged (...)

   The growth of the privileged class is accompanied by that of those holding higher degrees. Indeed, a general tendency of an increase in educational level, already observed previously, is particularly linked to the demographic growth of the privileged class. In all the spatial categories considered, the qualification of workers rose rapidly during the 1990s, a phenomenon that corresponds to the beginning of Luxembourg’s metropolisation phase and the beginning of the increase in privileged cross-border workers.5 One of the spatial categories typical of this phenomenon is that of Thionville, where the number of degree-holders increased the most in relation to other levels of education, and where the demographic difference between privileged and disadvantaged classes was the slightest. Metz is the most extreme spatial category in this respect: its demographic growth in border-crossers was the slowest, because it was, contrary to other socio-spatial categories (‘constant’ and ‘current hosting communes’), predominately the privileged classes that established themselves there – that is, the sociological profiles of border-crossers who remained inferior in raw numbers.

2. Beginning in 2008, the tendency of the privileged class to grow continues, compared with the relative stability of the disadvantaged class. This tendency shows that, beyond the limits which we formulate concerning the prominence of those with socio-professional status as executives and managers in the processes of metropolisation, this socio-dynamic has remained no less real, especially during the period which followed the financial crisis of 2008. Longwy remains, however, the only spatial category to maintain numbers which increase for both the privileged and disadvantaged classes of cross-border workers. We will return at greater length to this spatial category.

3. With the exception of Thionville and Metz, the proportion and the numbers of the disadvantaged classes remain higher than those of the privileged classes. Moreover, the gap in raw numbers between these two classes tends to increase, although, by contrast (figure 4), the proportion of the working-class falls and that of less-qualified employees follows the curve of the privileged classes (intermediate classes and managers or intellectual professions). In other words, the first part of our second hypothesis (2a) is only partially verified empirically. Indeed, the border-crossing workers from the privileged classes do not reside massively where the border-crossers from the disadvantaged classes were formerly most numerous. The category of ‘constant hosting communes’ shows this clearly, in that the numerical gap between the two classes has been increasing since the 1990s, while, let us recall once more, the portion of border-crossers from the disadvantaged classes diminishes. Thus, in those communes which have always hosted border-crossers, the portion of the disadvantaged social classes diminishes, but their raw number continues to grow more quickly than for the privileged classes. We observe the same tendency for Longwy, which is a spatial category that has long hosted cross-border workers.

4. Thionville and Metz are two spatial categories whose demography with respect to cross-border workers is particular: they are both those where the degree-holders are most numerous and those which host as many (Thionville) privileged as disadvantaged border-crossers, or even more (Metz). This result confirms the second part of the second hypothesis (2b): the more privileged social classes mainly take up residence in the secondary urban centres of the metropolitan area. This fact, however, is conditioned by an important temporal variable. Actually, it is only for these two secondary urban centres, which hosted resident cross-border workers only in the early stages of the metropolisation of Luxembourg (the 1990s), that we observe this demographic phenomenon. Indeed, in contrast to Longwy, Thionville and Metz were late in housing border-crossers going to Luxembourg. Moreover, Metz is the spatial category most typical of this socio-spatial phenomenon (cross-border demography initiated in the 1990s and number from the privileged classes higher than that of the disadvantaged classes), while Thionville indeed began late, but 10 years earlier than Metz, this process of residentialisation of border-crossing workers, and this beginning with the most disadvantaged social classes. Consequently, in Thionville, the latter have a raw number of cross-border works comparable to that of the privileged classes. In other words, the metropolisation of Luxembourg has had a tendency to include Thionville in a process of gentrification, while Metz has followed a process of the kind that metropolisation is seen to bring with it.

5. The agglomeration of Longwy remains that which incorporates the greatest number from the disadvantaged categories, compared with the other secondary urban centres. And it is also the only urban centre to have welcomed the cross-border workers from the most disadvantaged classes from the beginning, on the model of the ‘constant hosting communes’. Let us note as well that, like the latter, it is the urban centre closest to the frontier of the territory under study. This long-standing working-class social trajectory stems from the fact that this agglomeration was in large part the place of residence of Lorraine frontier-dwellers who worked in the steel industry before the economic transformation and the Luxembourg metropolisation. Thus it is currently one of the spatial categories where, on the one hand, the disadvantaged classes of cross-border workers are distinctly more numerous than their more privileged counterparts and, on the other hand, the raw number of the disadvantaged classes continues to grow. Finally, Longwy has a demographic trajectory among cross-border workers which is closer to the ‘constant hosting communes’ than to the secondary urban centres of the metropolitan area. Taken together, these facts enable us empirically to validate our third hypothesis, namely that the processes of residential segregations associated with metropolisation nevertheless depend in part on those which prevailed previously. The agglomeration of Longy clearly illustrates this socio-spatial process.

6. The ‘constant hosting communes’ reinforce the confirmation of our third hypothesis. Indeed, in comparison with the ‘current hosting communes’, they are (by virtue of their construction)) the communes which have taken in cross-border workers for the longest time. Now, since the first cross-border workers came from disadvantaged social classes, this socio-spatial precedence is now accompanied by a greater gap between the two social classes, to the detriment of the more privileged social classes (the more disadvantaged having comparable numbers as regards the two classes). In other words, as with the secondary urban centres, the rural communes experience a process of gentrification which is less intensive when they have already hosted disadvantaged cross-border workers before the metropolisation of Luxembourg. The level of education confirms this tendency, as has already been seen with the secondary urban centres (Longwy vs Metz and Thionville).

Conclusion and perspectives

28The results of this first quantitative analysis of the evolution of socio-spatial residentialisation logics of French-Luxembourg cross-border workers show clearly that the proportional decline of salaried employees from the disadvantaged (or working) classes confirms the process of cross-border metropolisation of Luxembourg [Sohn, 2012]. But they also show, as our first and principal hypothesis proposed, that this working-class fraction of cross-border commuters nevertheless still participates predominantly in this socio-spatial process, in proportion but also in absolute value (their number has not ceased to increase, although at a much diminished rate during this last decade), whereas numerous studies of metropolisation tend to obscure this. Moreover, when such studies deal with the working-classes it is generally to consider them as being relegated to the outskirts of the metropolis. Now, our results show that the situation is more complex, for, on the one hand, metropolisation does not create more socio-spatial fractures than the process of urbanisation which preceded it. On the other hand, the preceding socio-spatial distributions adapt the current socio-spatial reconfiguration, with certain spaces such as Thionville and Metz undergoing rapid social transformations since the beginning of Luxembourg metropolisation (1990s), while others, like Longwy, or the communes next to the frontier, are implicated in a more gradual gentrification. Thus the two broad socio-spatial dynamics formulated in the second part of our second (2b) hypothesis, and in our third (that, on the one hand, the role of the secondary urban centres in hosting the more privileged social classes, and, on the other, the effect of previous social segregations on segregations specifically metropolitan) seem to interact in the socio-spatial organisation of this metropolisation, without our knowing, however, if these phenomena are specific to our territory. Indeed, all the spatial categories having hosted cross-border workers from the beginning show a tendency to take in fewer who are typical of metropolisation (managers and intellectual professions with multiple degrees), because that first group was overwhelmingly working class. Thus the first part of our second hypothesis (that the spatial categories formerly reserved for cross-border workers from the working classes now receive those more qualified) is only partially verified, since it is precisely in these residential spaces (Longwy and the ‘constant hosting communes’) that the social transformation of their population is slowest. Similarly, the secondary urban centres of the metropolitan area have a role as residential reservoirs of the most qualified cross-border commuters, especially if their pasts allow it. Nevertheless, the diversity of the extent, as well as the pace, of urban gentrifications does not cast doubt on this general tendency. Thus the processes of gentrification of French city centres meets with continuity – is indeed extended to small urban centres like Thionville – by way of metropolisation, rendering socio-spatial partitions particularly complex, even if certain general tendencies remain valid.

30Thus these initial results remain essentially centred on the cross-border workers, even if certain structural and temporal aspects have been taken up and compared with residents of Lorraine working in France. Now, to what degree is the gentrification of the secondary city-centres of the metropolitan area dependent on socio-spatial processes that go well beyond Luxembourg metropolisation? Furthermore, in this process of metropolisation, what is the place of foreign immigrant manpower, at first situated outside the frontier regions and establishing itself essentially in Luxembourg: is it in competition with the manpower of Lorraine?