1French presidential elections periodically provide the opportunity to sketch the major political forces in the country via maps. These political forces obviously include the relation between the left and the right, symbolised by the map of the second round of the elections (excepting those of 1969 and 2002) ; but also the longevity of the political movements that get close to, and sometimes obtain, a “double-figure” score (the Front National on the extreme-right, as well as the centre, and the far left ; and also the “minor candidates” who rarely obtain more than 5% but whose geographical distribution often reveals deeper cleavages liable to become decisive in more favourable elections (representing political environmentalism, the extreme left, sovereignty etc.). This snapshot of French political opinion is all the more remarkable for the fact that it goes hand in hand with extremely high voter turnout (with a slight downturn in the second round of the 1969 elections), notably linked to intense electoral campaigning, almost exclusively transmitted via national media. The maps that result from these presidential elections can therefore be considered a relatively complete observatory of French political opinion. [1]
2The Ministry of the Interior has provided electoral data in digital form since 2002, which allows for mapping at scales that were hitherto impossible : covering several thousand cantons, and tens of thousands of communes. [2] The analytical methods of electoral geography, when applied to this very rich data, enable us to better understand electoral behaviour and to identify the evolutions and variables at work. We will make use of three tools in turn in this article. Firstly, we will draw on static cartography at the cantonal voting level for the major candidates in 2012 ; but also, in a dynamic perspective, we will map the evolution of political families, taking into account different reference periods, to reveal a detailed picture of what has changed and what remains the same in the French electoral landscape.
3Secondly, and this will be useful in explaining observations made previously, we will analyse electoral results in relation to census data, still at the level of the canton. The approach aims to show that the structure and weight of different socio-professional categories do not have the same impact everywhere on the vote ; or in other words, that social classes do not necessarily vote the same way in all areas.
4Finally, in the third section, we will analyse the results of the vote using another tool, the “urban gradient”, in order to explore a spatial dimension which is increasingly important in structuring the vote : the distance from major urban centres.
The geography of voting : a slow evolution which stabilised in 2012
5Rather than signalling a change, the maps of the 2012 presidential election appear to suggest continuity with previous elections. [3] Although the scores of Marine Le Pen or Jean-Luc Mélenchon exceeded those of their parties (the Front National and the Front de gauche, respectively) in previous elections, and were presented by the media as signalling a major change in the political equilibrium, the geographic structure of their electorates is far from new. Thus the vote for Marine Le Pen in 2012, calculated at the cantonal level, has a 0.95 correlation with Jean-Marie Le Pen’s score in 2007 ; it is difficult to see in this a geographical rupture comparable to the change in tone of the FN candidate, or even to see a clear spatial enlargement of the party’s spheres of influence. The vote for the Front de gauche, although significantly larger than that for Marie-Georges Buffet, the communist candidate in 2007, did not radically alter the electoral geography of the far left (there is a 0.74 correlation between the vote for Mélenchon in 2012 and the vote for Buffet in 2007). The relationship between votes for François Hollande in the first round, and the spatial distribution of votes for Ségolène Royal, the socialist candidate in 2007, is equally clear – the correlation here is 0.92, the same as for Nicolas Sarkozy in 2012 and 2007 (0.92). Changing candidates thus only has a marginal impact on the geographical implantation of the parties. Although their influence on the final score of their parties is considerable, they have little impact on the disposition of the party strongholds and battleground seats at the national level.
6The 2012 election will be analysed over two periods. We will begin by looking at the short-term, focusing on the picture painted by the 2012 maps and the changes between 2007 and 2012. Then we will look at the longer-term perspective of the last 30 years, and the illuminating comparison between “left-wing France” in 1981 and in 2012.
The differentiated erosion of Sarkozy’s electorate
7From Map 1, we can see that the decline recorded at national level in the vote for Nicolas Sarkozy (-4 points from the first round of the 2007 election) is reflected throughout almost all of the country, but is unequally distributed. Particularly important falls (between -5 and -13 points from 2007) occurred in the following five major zones : the Mediterranean coast and Corsica, Rhône-Alpes, the Massif Central, the inner East, and the Parisian basin extending to Picardie, Champagne-Ardenne and the Beauce.
The evolution of the vote for Nicolas Sarkozy between 2007 and 2012 in the first round of the presidential elections (as % of votes cast)
The evolution of the vote for Nicolas Sarkozy between 2007 and 2012 in the first round of the presidential elections (as % of votes cast)
8Most of the areas where the UMP (Union pour un mouvement populaire) candidate’s score declined the most significantly were areas where Marine Le Pen made substantial gains [4] and where this slide was not compensated for by the support of the centrist voters from 2007. This was the case along the Mediterranean coast and in Corsica, in the south-east of the Rhône Alpes region (Isère, Drôme), the inner East, Picardie and around the edges of the Parisian basin. Inversely, many of the historical stronghold areas of the Christian democrats, where the vote for the centrist candidate François Bayrou was severely eroded between 2007 [5] and 2012, represented limited losses for Nicolas Sarkozy.
9Although the vote for the FN did not take off in the “great Chiraquie” (the bastions of support for Jacques Chirac : Corrèze, the northern parts of the Lot and the Dordogne, the south of the Creuse and the east of the Cantal), the significant decline in Sarkozy’s scores in these areas can be explained by Hollande’s substantial gain in his stronghold over Royal’s score in 2007. In these areas, the socialist candidate managed to attract an electorate outside his traditional political camp, much like Chirac himself.
10However, the decline in the Sarkozy vote was more limited – a fall of 3 points at most – along the whole of the west coast of the country, in Paris and in the west of greater Paris, as well as in the north of Alsace. Other areas were also marked by stability or slight erosion of the Sarkozy vote, but these don’t appear on the map because they are wealthy but isolated communes. Whether in the provinces or around Paris, the areas where the bourgeoisie have their primary or secondary residences remained faithful to Sarkozy, and once again overwhelmingly supported him in the first round of the election. This magnifying effect, a result of focusing on communes of a certain size characterised by high income populations, [6] allows us to see the strong mobilisation on the part of the wealthy right-wing electorate in favour of the UMP candidate.
The sites of socialist progress
11The Corrèze area emerges clearly on Map 2 as the main centre of progression for the socialist candidate compared with Royal’s score in 2007. Other areas were also marked by significant gains in votes for the PS (Parti socialiste). Less compact and more dispersed, these areas correspond to the places where the vote for Bayrou was substantial in 2007, given that part of his electorate swung to the PS (or rather swung back, since the leader of the Modem party had attracted a substantial number of centre-left voters in 2007).
The evolution of the vote for the socialist candidate between 2007 and 2013, in the first round of the presidential election (as%of votes cast)
The evolution of the vote for the socialist candidate between 2007 and 2013, in the first round of the presidential election (as%of votes cast)
12This phenomenon essentially concerns zones situated along the western coast of the country : the Pyrenées-Atlantiques, Finistère, Anjou, the bocages [7] of the Vendée, and further north, the tip of Cotentin.
13The progression of the socialists was also significant in Ile-de-France and in Haute Normandie, but also in Picardie, Nord-Pas-de-Calais and Champagne-Ardenne. In this latter region they obtained 35% of the vote (+6%) in Charleville-Mézières, a symbolic town for Sarkozy where he delivered his famous speech on the value of work, aimed at “the France that gets up early” (“la France qui se lève tôt”), and a proportion of this electorate was receptive to this rhetoric, particularly in these regions of the inner East. Although the efforts to win back industrial working-class French voters paid off in these regions, the results were more limited in Lorraine, Alsace, Franche-Comté and in the northern half of the Rhône-Alpes region (excepting the Lyon metropolitan area). In the south, a series of declines were recorded in cantons in Isère, the Drôme, the Southern Alps, and even the Var. In these areas (where Arnaud Montebourg had obtained strong scores in the socialist primaries), the decline of the socialist vote was accompanied by gains made by the Front de gauche. The same phenomena can be observed on the other side of the Rhône valley, in Ardèche, in the Cévennes, and even further west, in Corbières, Ariège, Haute-Garonne, Tarn-et-Garonne and part of the Landes. In these rural and/or mountain regions, particularly sensitive to issues relating to the preservation of public services and with a strong socialist and radical tradition, the substantial support which Jean-Luc Mélenchon’s discourse found was stiff competition for François Hollande.
The erosion of the historic structure of the centrist vote
14In analysing the evolution of the vote for the PS and the UMP between 2007 and 2012, on several occasions we have implicitly come up against another major political event of the 2012 election : the significant haemorrhaging of support for Bayrou, which saw his score drop from 18.6% in 2007 to just 9.1% in 2012. Almost mechanically, it was in those areas where support for Bayrou had been highest that the losses were largest. No area was exempt from this (which means that all elements of Bayrou’s electorate were affected by these centrifugal movements). The map of the Modem vote between 2007 and 2012 (Map 3) corresponds in all points with the map of the historical structure of the Christian democrats combined with Bayrou’s own local political base.
The evolution of the vote for Bayrou between 2007 and 2012
The evolution of the vote for Bayrou between 2007 and 2012
The vote for Mélenchon : what was new and what remained the same
15The map of the vote for Mélenchon is broadly similar to the historical and traditional geography of the communist vote. Many of the Front de gauche candidate’s zones of strength correspond to communist strongholds. [8]
16This was, for example, the case in the Nord-Pas-de-Calais region, as well as in the mining basin and Valenciennes ; in Seine-Maritime ; in Ile-de-France with the communes of the “red suburbs” ; in the East with the region of Longwy ; but also in the Gard, or the communist strongholds of the Bouches-du-Rhône. A significant vote for Mélenchon can also be observed in areas where there is a strong tradition of rural communism, for example in the centre of Brittany (the Monts d’Arrée) or in the bocage Bourbonnais area in Allier.
17But the success of Jean-Luc Mélenchon’s candidacy lay also in his ability to attract electors from the left beyond the ranks of communist sympathisers. If the vote for the Front de gauche was relatively high and homogenous in the southern half of France, the situation was much more contrasted north of the Loire river. Here there was a vast amount of terrain to be conquered, from the strongly conservative areas in the East, to the traditional zones of the right (Beauce, Perche, the bocages of Normandy and the Vendée) in the western part of the country. The Front de gauche nevertheless managed to find support in most of the major urban areas in the west, where Jean-Luc Mélenchon obtained respectable scores in the town centres and their immediate peripheries. These towns in the west constituted the principle strongholds of the PS in that region and the development of the Front de Gauche vote there was achieved on the back of an electorate that was mainly socialist (with the support of votes from the extreme left). The difference with the surge in the vote in the south lies in the fact that here it essentially concerns an urban socialist electorate, whereas it affected rural cantons as well as towns in the south-east and south-west. Given its foothold in urban areas, certain commentators described the vote for Mélenchon as a “bobo” [9] vote. The candidate of the Front de gauche did indeed attract high scores in hyper-urban areas where the “bobo” lifestyle is common. This was the case in the eastern parts of Paris : 17.4% in the 20th arrondissement, 15.7% in the 19th, 15.3% in the 18th, and 14.9% in the 10th along the Canal Saint-Martin.
18The same phenomena can be observed in the 1st arrondissement of Lyon (19.9%) or in Grenoble (15.4%), a city where the Greens regularly achieve high scores. However, reading Map 4 we can see that, in the end, the gentrified [10] town centres only represent a small proportion of the Front de gauche’s areas of strength ; the urban working-class towns, especially those with a strong communist tradition, and the rural cantons weigh more heavily in that support.
The vote for Jean-Luc Mélenchon in the first round of the 2012 presidential election (as % of votes cast)
The vote for Jean-Luc Mélenchon in the first round of the 2012 presidential election (as % of votes cast)
Left-wing France… 30 years on
19The fact that François Hollande won the presidential election with almost exactly the same score as François Mitterrand in 1981 (51.6% compared to 51.8% then) lead to comments on the unchanging nature of the struggle for power between the left and the right in France. However, if we look more closely, we can see profound changes behind this apparent stability. In several regions, spectacular evolutions have occurred which have led to a marked reconfiguration of the French political landscape. The map comparing the vote for Hollande with the vote for Mitterrand [11] 31 years earlier speaks volumes in this respect. Indeed, it shows a double movement : a clear progression of the left in the west of France, and inversely, a significant erosion of its positions in the east and the south-east.
20Although the progress of the left in the west includes most of Basse-Normandie, the Pays-de-la-Loire, Brittany, [12] Poitou-Charentes, and the Pyrénées-Atlantiques, it was particularly strong (up at least 7 points) in the backbone of the inner west (the bocages of the Manche, Ille-et-Vilaine, the west of Mayenne, the countryside of the Loire-Atlantique and the east of Morbihan, Chotelais, and the bocages of the Vendée), as well as in most of Finistère, the Deux-Sèvres and the Pyrénées-Atlantiques. In all these very rural and historically very Catholic areas, urbanisation and the decline of agriculture and religious practice have allowed the left to make progress since the beginning of the 1980s. If religion has lost ground in these regions, Catholic culture continues to indirectly influence local sensibilities, which are more in favour of moderate and pro-European political discourses. It was in this context that the PS was able to significantly increase its audience with a more centrist discourse than at the end of the 1970s and beginning of the 1980s (abandoning references to Marxism and class struggle, the end of the “school wars” etc.), whilst Sarkozy’s discourse, which was much tougher than that of the right at the time on issues such as the construction of Europe, immigration and security, seemed to persuade fewer voters in these areas. The same shift in favour of the left (combining long-term socio-cultural evolutions and more recent ideological choices) can be seen in other rural regions with the same Catholic heritage : Cantal, Haute-Loire and part of Aveyron and Lozère to the south of the Massif Central, and to a lesser extent in the Pays de Caux, Flanders or in Moselle.
21But the comparison with 1981 is not entirely advantageous for the left. Although Hollande managed to significantly improve the socialists’ score in western France compared to his predecessor, he also recorded spectacular declines in some areas, particularly in the departments on the Mediterranean.
1981-2012 : a comparison of results of the left in the second round of the presidential elections in the departments of the west and south-west (%)
1981-2012 : a comparison of results of the left in the second round of the presidential elections in the departments of the west and south-west (%)
The difference between the vote for Mitterrand in 1981 and the vote for Hollande in 2012 in the second round of the presidential elections
The difference between the vote for Mitterrand in 1981 and the vote for Hollande in 2012 in the second round of the presidential elections
22For at least twenty years, the departments of the PACA region [13] have been presented as a stronghold of the right, but the table above reminds us that in 1981 François Mitterrand overwhelmingly won the vote in the Bouches-du-Rhône and in Vaucluse ; that he was only just beaten in the Var, and a little more comfortably beaten in the Alpes Maritimes. Thirty years later the situation is radically different and it is now the right that reigns uncontested, the left having lost between 9 and 11 points on average depending on the department. Although the struggle for power is more balanced in Languedoc-Roussillon, François Hollande still received a very low score compared to that of François Mitterrand in the “Red Midi” (“Midi Rouge”). Along the Mediterranean coastline, certain socio-demographic changes (decline of the working rural population and the local working class, the emergence of the service sector and the tourism industry, and the influx of retirees), have progressively sapped the electoral foundations of the left leaving it a minority party in these areas.
23This decline of the left can be seen to a lesser extent, and in a slightly different way, in the Garonne valley, but also in many industrial areas in eastern France (the Sochaux-Montbéliard basin, the eastern region around Lyon, the Saint-Dizier basin and that of Vitry-le-François, the metal-working areas in Lorraine, and the area around Valenciennes etc.). The decline or even the disappearance of whole areas of industrial activity has modified the political sociology and behaviour of these regions. This was the case in departments such as Seine-et-Marne, Marne, Aube, Oise or even the Haute-Marne (in the latter two departments, François Hollande received a minority share of the vote, and trailed Mitterrand, who was in the lead there in 1981, by 7.3 and 6.4 points respectively).
Contrasting effects of the social structure on the vote in different regions
24Major regional changes are of course linked, in part at least, to the major economic changes in the territories discussed above. It is common, however, to suggest that the “heavy” explicative variables of the vote (particularly socio-professional status) count for less today than they used to in terms of electoral choice. Combining the distribution of the vote in 2012, at the cantonal level, and the distribution of census data [14] in the same geographic information system, enables us to calculate the correlations between political and sociological variables. It is important, however, to avoid falling into the trap of the ecological fallacy, in other words deducing individual behaviour from collective averages. Thus a high correlation between the number of workers in an area and the vote for the left does not signify that workers vote left, but simply that they are more numerous than other socio-professional categories in areas where people tend to vote left. The use of variables such as the number of workers, managers, employees, foreigners, or university-educated people in an area is not intended to identify individual behaviour but instead to take these variables as markers of territorial configurations. [15] Thus the level of workers in an area would be considered as a relevant indicator of those cantons characterised by industrial activity ; or the level of managers or university-educated people might be representative of cantons with service sector industries.
Heavy variables which remain decisive
25The calculation of correlations based on c.4000 French cantons between six candidates (as well as the left/right balance in the first round) and eight social variables, never leads to particularly high scores (correlations can theoretically vary between -1 and +1). This seems logical given the high number of spatial units used, knowing that correlations generally increase inversely to the number of statistical units used to calculate the correlation.
26However, these correlations are far from random. The vote for Le Pen is the most strongly correlated to the presence of workers (0.48), well above that of Mélenchon (-0.17) or Hollande (-0.16). The strong presence of workers in a canton is thus a good predictor of a strong vote for the FN, more than a vote for the traditional left, and even more than for the Greens (-0.45). The vote for Le Pen is however negatively correlated with the presence of university-educated populations and managers, which broadly corresponds to the party’s weak foothold in town centres that have a substantial proportion of these more affluent social categories. It is important to note that the correlations which measure the links between the distribution of socio-demographic variables and the distribution of the vote for the FN candidates are almost the complete opposite of those which measure the same links with the distribution of votes for Eva Joly.
Correlations between socio-demographic variables (2009 census) and votes (% of enrolled voters) for the main candidates at the 2012 presidential elections, at the level of French cantons
Correlations between socio-demographic variables (2009 census) and votes (% of enrolled voters) for the main candidates at the 2012 presidential elections, at the level of French cantons
27The left’s score (Mélenchon and Hollande) proved “indifferent” to the strong presence of managers, workers or university-educated populations, but on the other hand there was a closer correlation with the presence of employees as well as with the unemployment rate, exceeding even the Le Pen vote on this criteria. The most favourable cantons for the socialists appear to be socially “intermediate” areas, between the industrial basins in crisis (where the FN has made advances), and the more residential communes which remain the established province of the right. The positive correlation between the left and the rate of social housing, in places where the right shows negative correlations, confirms this observation.
28Finally the correlation between the presence of foreigners and the vote for Le Pen is negative (-0.20) at cantonal level. In fact it is only rarely significant, and when it is, it is positive for the overall left (0.30). This statistical connection may be linked to the socialist score in the banlieues of the major urban centres.
29This analysis does not allow us to draw conclusions as to individual behaviour (which has been studied elsewhere in national surveys) ; it only reveals a virtual national reality (as other studies do) which is rarely verified on the ground. The advantage of these national correlations is thus that they may be compared to other, similar but more contextualised, calculations, which is difficult to achieve in national surveys.
Contextual effects
30Several previous analyses have demonstrated that these national correlations may disappear, or even be inversed, when they are measured at the level of a department or a region rather than at national level. [16] We have calculated the correlations between the main candidates in the presidential election and two socio-economic variables (the level of managers and workers) in six departments or groups of departments. The choice – which is of course simplistic – of an analysis based simply on the level of workers/managers in an area, aims above all to demonstrate the diversity of politico-economic configurations in France. It is worth remembering that we seek to use these indicators to shed light on the industrial areas, as distinct from service sector areas – this being, over and beyond the distinction between working-class employees and managers, the major economic cleavage throughout France.
Correlations between two social variables from the census (2009 data) and the main candidates in the 2012 presidential election, at cantonal level within certain French departments
Correlations between two social variables from the census (2009 data) and the main candidates in the 2012 presidential election, at cantonal level within certain French departments
The inner suburbs around Paris (La petite couronne) : a left/right, workers/managers opposition
31The inner ring of suburbs around Paris (Hauts-de-Seine, Val-de-Marne and Seine-Saint-Denis) is typical of the decisive nature of the variables we have chosen to use. The strong presence of workers is very positively correlated to a vote for the left (0.7), whereas the vote for the right is highly correlated to the presence of managers (0.81). The values of the correlations are also much higher than at the national level. They illustrate the residential and economic segregation that is decisive in electoral decisions, and which are measured here in the “petite couronne” around Paris, but which appear to be characteristic of the functioning of the major urban centres where a majority of French people live.
North and south-west : working-class cantons won over by the Front National
32The configuration is not the same in the Nord department in the north of France. Although the correlation between the vote for the right and the percentage of managers remains positive, and the correlation between the vote for the right and the percentage of workers remains very negative, the score of the left (both Mélenchon and Hollande) is low in the working-class cantons ; conversely there is a very strong correlation recorded in these cantons with the Le Pen vote (0.76).
33In the different context of the south-west, a rural area very favourable for the socialists, the correlations appear similar. There is a positive link between the vote for the FN candidate and working-class cantons, and between the right-wing vote and the presence of managers, and a negative link between the left-wing vote and the presence of managers. But unlike in the Nord, in the south-west, the level of workers is not negatively correlated with the right (-0.04) but is more so with the vote for the left (-0.17). Clearly in the south-west, and to a lesser extent in the Nord, the vote for the left does not seem to be, or not to be any longer, structured around the industrial strongholds ; instead it has been supplanted on this terrain by the Front National, whereas the right remains polarised in the more affluent areas.
Vendée, Var and Alsace : reverse models
34Regional contexts can inverse these correlations however, and help us to relativize them. Thus, although the vote for Le Pen is almost always positively correlated with the presence of workers, this is not the case in the Vendée. The rural working-class cantons of one of the most industrial departments in France (although it is rarely perceived as such) are characterised by a very low vote for Le Pen (-0.31), such that this might be termed the “Vendée paradox”.
35In the Var, the right is not very strongly correlated to the presence of managers (-0.10), although it is still negatively correlated to the presence of workers (-0.49). Globally these correlations are not very significant. The case of Alsace is more so, where although working-class cantons vote heavily for Le Pen (0.67), the socialist left is very negatively correlated with the rate of workers (-0.45) and is positively correlated with the presence of managers (+0.20).
36These few examples demonstrate that, although global trends may exist which enable us to link electoral behaviour to strong socio-economic configurations, particularly in an urban context, these trends do not necessarily hold when they are studied only in certain areas (Vendée, Alsace, etc.). We should therefore seek an electoral explanation via a combination of global explanatory models that take into account local configurations. This is a major contribution to the geographical approach to understanding the vote. It is not about looking for a “geographical optimum” or one scale of analysis that would be more pertinent than another (seeking the most detailed possible dissection of the vote, for example), but instead about defining a combination of scales. [17]
The contribution of the urban gradient in understanding the vote
37Beyond electoral maps and the possible comparison with the socio-economic landscapes of France, the distribution of the vote allows for a third possible analysis : spatial analysis, [18] as it is known by geographers. Here we are no longer simply concerned with varying scales, as in the two previous sections, but with measuring the effect of spatial interactions, or in other words, calculating whether the score of one party in a given area is influenced by the score of that party in a neighbouring area. These measures may be based on indicators of distance, such as the density of the vote for each party based on distance from urban centres. [19]
The dynamic of Marine Le Pen
38Although the left/right cleavage continues to be more or less organised around the major regions, particularly with an east-west division, take into account where people live, and particularly the distance from the nearest urban centre, is equally important in understanding the vote. Over the course of the last 30 years the social geography of France has been profoundly modified by urban sprawl and peri-urbanisation. Entire sections of the working-class have left the major urban centres to move to peri-urban residential areas, as part of a phenomenon of relegation that they either submit to (increases in housing prices in city centres lead them to move or to buy houses in more affordable zones ever further from the centre), or choose (escaping from the big city, the suburbs and their perceived nuisances). [20] Parallel to this, the major urban centres have seen an increase in immigrant populations in working-class areas, along with the concentration of large numbers of managers and intellectual professionals. In this new social landscape the “urban gradient”, that is the distance between where people live and the closest major urban centre (with more than 200,000 inhabitants) has an increasingly significant electoral impact. This is clearly the case for the vote for the Front National, which is very much structured on this principle.
39In the first round of the presidential election we thus observed (see Figure 1 below) an “excess-vote” for the FN and a very significant breakthrough in the greater peri-urban areas [21] (which combine rural areas and small towns situated between 20 and 50 km from the urban centres). Inversely, the metropolitan areas recorded scores for Marine Le Pen that were much lower than average, and falling. Everything seems to suggest that this year the structure of the vote for the FN was even more polarised along spatial lines than in the past.
Comparison of the vote for the FN in the first round of the presidential election from 1995 to 2012, according to the distance from urban centres of more than 200,000 inhabitants
Comparison of the vote for the FN in the first round of the presidential election from 1995 to 2012, according to the distance from urban centres of more than 200,000 inhabitants
40More than ever, the heart of the major urban centres stands out as a space of resistance to the vote for Le Pen [22] (14.8% in 2012, down 1.5% from 2002), which instead prospers more and more in the greater peri-urban areas. In places situated between 30 and 50 km from urban centres the level of the vote for Marine Le Pen stabilised on average at around 21%, whereas her father had obtained “only” 18% in 2002 and 16.5% in 1995. The ongoing development and embedding of the FN vote in peri-urban areas since the 1995 presidential election (with the exception of the atypical election of 2007) can also be seen in the rural areas that are even further away.
1995-2012 : the evolution of the vote for the FN in the first round of the presidential election in rural France (%)
1995-2012 : the evolution of the vote for the FN in the first round of the presidential election in rural France (%)
41Although the progress is quite spectacular in rural areas over the long term (we will come back to this), Figure 1 nevertheless shows that it is in the greater peri-urban area that the level of the vote for the FN is the highest, in 2012 as in previous elections. As a result it is important to interrogate the origin of this “excess-vote” for the FN in the larger peri-urban areas. Can it be explained by structural effects ? That is, by the fact that the working-class populations (who make up the categories who have voted most for Marine Le Pen) are over-represented in these areas ? Or is the vote for Marine Le Pen particularly pronounced among the workers and employees in these zones ?
42First, according to the Ifop survey conducted on the day of the first round of the election, [23] Marine Le Pen obtained 33% of the vote amongst workers and 28% amongst employees. It was in these two groups that she obtained her best results (by comparison, she only reached 13% amongst self-employed and senior management groups for example) and topped the polls for these two groups (admittedly, on equal footing with François Hollande amongst employees). The FN’s strong following in working-class areas, measured with objective categories, is also apparent if we consider subjective categories. Still using the same survey, the FN candidate’s score perfectly correlates with the difficulties respondents encounter in making ends meet with their household income : 7% amongst those who make do “very easily”, compared to 32% amongst those who have “great difficulty” making do.
The vote for Marine Le Pen indexed according to the ability to make do with one’s income
The vote for Marine Le Pen indexed according to the ability to make do with one’s income
43It is also clear that among those who say they have “great difficulty” making do with their income it is Marine Le Pen who comes first with 32%, ahead of François Hollande (24%) or Jean-Luc Mélenchon and Nicolas Sarkozy (15% each). Working-class groups and those experiencing difficulties do indeed constitute the primary element of the vote for the FN at the national level. Yet a detailed analysis shows that the areas that we have called the greater peri-urban areas (including both peri-urban communes and also small towns or rural areas situated between 30 and 50 km from a major urban centre) are characterised both by the higher proportion of workers and employees at all levels of our urban gradient, and by a vote for the FN that is higher than average in these working-class groups : 32% of workers and employees living in a 30 to 40 km radius and 34.9% of those living in the next level of the gradient (40 to 50 km) voted for Marine Le Pen, compared to just 28.9% on average.
The importance of working-class categories and their voting rates in favour of Marine Le Pen for communes at different levels of the urban gradient (%)
The importance of working-class categories and their voting rates in favour of Marine Le Pen for communes at different levels of the urban gradient (%)
In order to have a sufficient statistical base for employees and workers, the last three levels, the least populated, have been grouped into one : + 80 km from a major urban centre.44It is thus the cumulated effect of these two factors that contributes to the fact that Marine Le Pen obtained her best results in these specific areas. The “excess-vote” of the working-class categories in favour of the FN reflects a certain number of difficulties in these areas : low income levels, fragile local economic fabric, the concentration of work in a small number of businesses (thus amplifying the impact in case of site closures, workers having more difficulty finding jobs than in urban centres), heavy dependency of these populations on the price of petrol, increases in minor delinquency and the disappearance of certain public services. Added to all this are the consequences of the relegation – both submitted to and chosen (the obligation or the choice to leave major cities) and the fear “of being caught up in the suburbs”, a worry that haunts a proportion of these working-class populations in these greater peri-urban areas. Although they often suffer from being too far away from big cities (significant cost and time spent on transport, fewer local services or lower quality services, which lead to feelings of being second-class citizens), they also fear that urban nuisances in the shape of delinquency or “problem groups” (immigrants or social welfare recipients) will progressively spread to them. [24]
45Marine Le Pen and her advisors have perfectly pinpointed the psychological profile of this working-class, peri-urban electorate. Indeed, it was often first to them, the famous “invisible people”, that she addressed her messages during the campaign, clearly calibrating her discourse to play to the expectations and daily difficulties of this electorate. Well researched and adapted to the psychological and social reality of this specific terrain, Marine Le Pen’s version of the message of the FN is in direct competition here with the message of the PS. Although François Hollande managed to take the lead against Le Pen in the working-class electorate in the major cities, as we can see in the table below, he lost ground in the inner suburbs and was clearly outrun among the employees and workers in the greater peri-urban area. The gap in favour of the FN candidate even approached 12 points in the communes situated in a 40-50 km radius from a major urban centre.
The vote in favour of François Holland and Marine Le Pen among workers and employees in communes at different levels of the urban gradient
The vote in favour of François Holland and Marine Le Pen among workers and employees in communes at different levels of the urban gradient
In order to have a sufficient statistical base for employees and workers, the last three levels, the least populated, have been grouped into one : + 80 km from a major urban centre.46Beyond the working-class categories we have been interested in so far, the following diagram (Figure 3) clearly shows that the “excess-vote” for the FN among peri-urban populations (working-class and other categories) is accompanied by scores for François Hollande that are noticeably lower than his national average, whilst those of Nicolas Sarkozy are stable. In this area (and particularly among the employees and workers who live there) a political and ideological battle pits the PS against the FN, with the latter undoubtedly benefiting from the mismatch of the socialist discourse in these areas.
Deviation from the national average of votes for the three main candidates according to the distance from major cities
Deviation from the national average of votes for the three main candidates according to the distance from major cities
47This diffusion phenomenon has also affected the rural territories where the FN had always encountered difficulties but now finds itself scoring at a much higher level : in the centre of France and in the Grand Ouest regions. In these areas, and in line with what we have observed in our analysis using the urban gradient, this shift concerns the peri-urban or even very rural areas, rather than the large or medium-sized towns that structure the territory. Another noteworthy element is that the advancement of the FN is occurring on the territory of both the left and the right. In Auvergne and in the Limousin, Marine Le Pen has significantly improved her father’s score in the second round of the presidential election of 2002 in the very Catholic and right-wing Cantal, but progress is just as significant in the “red” and “de-Christianised” countryside of Allier, Haute-Vienne, Puy-de-Dôme and the Creuse.
48Further north, these gains are also seen in zones that are just as rural but much more conservative. This is the case for example in what geographers [25] call the “backbone of the inner West” (la dorsale de l’Ouest intérieur), a vast zone of bocages running from south of the Manche to the Vendée, a very Catholic area that constitutes one of the most favourable areas for the traditional right in France.
The influence of cities on François Hollande’s victory
49The historic score of the left in Paris (55.6%) is not an isolated case, quite the contrary. Indeed, the geography of the vote for Hollande is also characterised by clear domination in most large cities, whether they are run by the left or the right.
A clear advantage for François Hollande in major cities (%)
A clear advantage for François Hollande in major cities (%)
50The personality and the message of the socialist candidate clearly resonated with the expectations and values of many of the inhabitants of the major cities in these large metropolitan areas (Hollande obtained 56% of the vote amongst senior managers and independent workers). [26] However it is important to stress that suburban France also made a significant contribution to his victory. Indeed, while Ségolène Royal had already obtained good scores in these suburban communes, François Hollande further amplified this shift, undoubtedly on the back of an anti-Sarkozyism that was more virulent in these towns than it had been five years previously.
In the communes of the suburbs, François Hollande improved the already high scores obtained by Ségolène Royal in the second round
In the communes of the suburbs, François Hollande improved the already high scores obtained by Ségolène Royal in the second round
51Although François Holland came out on top in the major urban centres and their immediate surroundings (in both the first and second rounds), the power struggle between the left and the right varied significantly according to the urban gradient, as we can see in Figure 5. The socialist candidate was clearly ahead of his opponent (55% against 45%) in areas situated less than 10 km from the centre of any of the twenty largest cities in France. The result was much closer, however, if we look at the next level of the gradient. François Hollande only obtained 51% of the vote for persons living between 10 and 20 km from a major city. Further away still, the greater peri-urban area appears to have been a battleground for the PS. The UMP candidate scored between 51% and 52% in these areas (situated between 20 and 50 km from a major city), compared to 49% to 48% for the socialist candidate, down no fewer than 6 or 7 points from the score in the city centres. The vote in rural areas which were further away (more than 70 km from a major city), covering both the right-leaning countryside but also many traditionally left-leaning areas (Land, Lot-et-Garonne, Corrèze, Lot, Nièvre etc.), gave a clear advantage back to François Hollande who obtained between 53 and 54% there. This also demonstrates that the measure of the distance from cities cannot be analysed without taking into account the significant regionalisation of the vote, particularly in rural areas.
Comparative results for François Hollande and Nicolas Sarkozy in the second round, according to the distance from major cities
Comparative results for François Hollande and Nicolas Sarkozy in the second round, according to the distance from major cities
52* * *
53These three illustrations of the contribution of electoral geography (detailed national mapping, socio-electoral correlations at infra-departmental levels, spatial analysis based on the distance from cities) do not by any means aim to provide an explanation of the act of voting and its political direction that would be sufficient in and of itself. Thus the differences in the correlations between parties and the main socio-professional classes does not reveal contextualised individual electoral attitudes (to do so would be to fall victim to the ecological fallacy). They do however, provide an indispensable counterpoint enabling us to relativize the explanatory contribution of national surveys, as refined as the latter might be. Similarly, the “urban gradient” must not be considered as a new “heavy” variable in explaining the vote. It is not in itself the distance from the city which explains the drop in the socialist vote or the increase in the vote for the FN, and it is still less the peri-urban residential environment that is in itself predictive of this “excess-vote” for Marine Le Pen. The distance from the city is only a “model”, used here to shed light on processes which appear to be fundamental : such as the increase in housing prices and the “exclusion” of certain parts of the middle classes from the centre of cities ; the crisis of individual car-based mobility because of travel costs and urban obstacles ; the absence of restrictive legislation in rural and peri-urban areas to encourage social diversity, or shared, rental or communal housing. Thus it will undoubtedly be difficult in the short term for the PS to make up its lack of votes in the greater peri-urban area without making substantial structural territorial reforms, likely to be unpopular with a majority of local representatives (for example drastic control of urban sprawl, implementation of a genuine supra-communal governance of urban centres, reducing prerogatives for mayors in terms of urbanism etc.).
54It is also important to bear in mind that the measures used here only reveal average deviations between different places of approximately a few percentage points. Therefore, mathematically, they only concern a minority of voters, even if that minority is important in understanding socio-electoral processes or decisive for the result of the vote in a local election. Thus, the majority of workers in peri-urban areas do not vote for the Front National.
55On a methodological level, our analysis also illustrates the progress made in the processing of electoral data in the last fifteen years, which now enables us to rapidly mobilise data after a national election, by canton and by commune, and to compare them with previous elections, or socio-economic data from recent censuses. Two further steps may be taken in future years. One is the analysis at an even closer level than that of cantons and communes – that of polling stations. [27] The other is the linking of these very granular divisions of data (polling stations in towns, communes in rural areas) with data from “geo-localised” studies enabling us to combine individual attitudes and effects of context, as we have done in the last part of this article (Tables 5 and 6). The disciplinary borders appear to be sufficiently open today to allow such a dialogue, both on a methodological level and in terms of a renewed analysis of the explanatory factors of the vote.
Notes
-
[1]
Michel Hastings, “Les démiurges de l’introspection cartographique. A propos de La France qui vote de Frédéric Bon et Jean-Paul Cheylan”, Politix, 2(5), 1989, 74-6.Online
-
[2]
Translator’s note : The commune is the lowest administrative level in France, roughly equivalent to a town, a city or a municipality. A canton is an electoral constituency for the election of conseilleurs généreux, members of the Conseil Général (general council) at the departmental level. There are generally many cantons per department (40 on average) and sometimes several within a single town. The département (department) is the second administrative level in France, after the régions which are made up of several departments.
-
[3]
See especially Pierre Martin, Comprendre les évolutions électorales (Paris : Presses de Sciences Po, 2000) ; Pierre Bréchon, Les élections présidentielles en France, Quarante ans d’histoire politique (Paris : La Documentation Française, 2007) ; Bruno Cautrès, Anne Muxel (eds) Comment les électeurs font leur choix ? Le Panel électoral français 2007 (Paris : Presses de Sciences Po, 2009) ; Michel Bussi, Jérôme Fourquet, “Neuf cartes pour comprendre l’élection présidentielle de 2007”, Revue française de science politique, 57(3-4), 2007, 411-28.
-
[4]
Often because of voters returning to the FN after having voted for Sarkozy in 2007. On Sarkozy’s electorate in 2007, see the edition of the RFSP dedicated to that presidential election. Revue française de science politique, 57 (3-4), 2007.
-
[5]
Sylvie Strudel, “La tentation bayrouiste”, in B. Cautrès, A. Muxel (eds) Comment les électeurs font leur choix ?…, 221-38.
-
[6]
This phenomenon can also be observed at a more micro level within major cities, if we drill down to the level of neighbourhood or polling stations. For more details see http://www.cartelec.net.
-
[7]
Translator’s note : a bocage is a type of landscape, a patchwork of fields bordered with hedgerows.
-
[8]
See for example, Marie-Claire Lavabre and François Platone, Que reste-t-il du PCF ? (Paris : Autrement, 2003).
-
[9]
Translator’s note : “Bobo” is an abbreviation of “bourgeois-bohème”, describing a lifestyle that is both affluent and alternative.
-
[10]
Anne Clerval, Antoine Fleury, “Politiques urbaines et gentrification. Une analyse critique à partir du cas de Paris”, L’Espace politique, 8(2), 2009, <http://espacepolitique.revues.org/index1314.html>. Online
-
[11]
Alain Lancelot, 1981, les élections de l’alternance (Paris : Presses de Sciences Po, 1986).
-
[12]
On the political evolution of the Breton electorate, see Jean-Jacques Monnier, Le comportement politique des Bretons (Rennes : Presses Universitaires de Rennes, 1994).Online
-
[13]
PACA is an abbreviation of Provence-Alpes-Côte-d’Azur.
-
[14]
From Insee the French national bureau of statistics.
-
[15]
Peter J. Taylor, Ron J. Johnston, Geography of Elections (Harmondsworth : Penguin Books, 1979), and more recently Nonna Mayer, Sociologie des comportements politiques (Paris : Armand Colin, 2010).
-
[16]
On the regional dimension of the vote, see Claude Dargent, “La notion de culture politique régionale est-elle pertinente aujourd’hui”, in Pascal Perrineau, Dominique Reynié (eds), Le vote incertain, les élections régionales de 1998 (Paris : Presses de Sciences Po, 1999), 45-69. See also, for example, Nonna Mayer, “De Passy à Barbès : deux visages du vote Le Pen à Paris”, Revue française de science politique, 37(6), 1987, 891-906.
-
[17]
John E. Agnew, “Mapping politics. How context counts in electoral geography”, Political Geography, 15(2), 1996, 129-45 ; Ron J. Johnston, Kelwyn Jones, Carol Propper, Rebecca Sarker, Simon Burgess, Anne Bolster, “Party support and neighborhood effect. Spatial polarisation of the British electorate, 1991-2001”, Political Geography, 23, 2004, 367-402.
-
[18]
Spatial analysis, frequently used by geographers, aims to theorise, measure and model spatial organisation, for example using the notions of distance, gravity, networks, location etc.
-
[19]
Loïc Ravenel, Pascal Buléon, Jérôme Fourquet, “Vote et gradient d’urbanité : les nouveaux territoires de l’élection présidentielle de 2002”, Espace, populations, sociétés, 21(3), 2003. 469-82 ; Michel Bussi, Céline Colange, Jean Rivière, “Distance(s) à la ville et comportements électoraux. Quelques éclairages quantitatifs lors des derniers scrutins présidentiels”, in Marie-Flore Mattéi, Denise Pumain (eds), Données urbaines 6 (Paris : Anthropos, 2011), 33-42.
-
[20]
See for example, Jacques Lévy, “Peri-urbain ; le choix n’est pas neutre”, Pouvoirs locaux, 56, 2003, 35-42.Online
-
[21]
Jean Rivière, “Le vote pavillionaire existe-il ? Comportements électoraux et positions sociales locales dans une commune rurale en cours de périurbanisation”, Politix, 83, 2008, 23-48 ; and “Trajectoires résidentielles et choix électoraux chez les couches moyennes périurbaines”, in “L’espace des classes moyennes”, Espaces et sociétés, 148-149, 2012, 73-90.
-
[22]
Pascal Perrineau, Le symptom Le Pen. Radiographie des électeurs du Front national (Paris : Fayard, 1997).
-
[23]
Ifop/Fiducial survey, for Europe 1, Paris Match and Public Sénat on 22 April 2012, with a nationally representative sample of 3509 people registered to vote.
-
[24]
The inhabitants of these areas see their relationship to the major cities as “too far away for the benefits, but close enough for the inconveniences”.
-
[25]
See Michel Bussi, Eléments de géographie électorale. A travers l’exemple de la France de l’Ouest (Mont-Saint-Aignan : Publications de l’Université de Rouen, 1998).
-
[26]
Ifop/Fiducial survey for Europe 1, Paris Match, and Public Sénat, 6 May 2012, with a nationally representative sample of 1968 people registered to vote.
-
[27]
See the website for details : <http://cartelec.net>
