Who is Responsible for the Disappearance of Social Classes?

The Decline of Occupational Classifications

In his presentation, Alain Desrosières reminds us that at the origins of the “CSPs” there was the work of Jean Porte, a specialist in mathematical logic who entered INSEE after World War II. In charge of the 1954 census, he was faced with a request for a classification, in a context where Marxism deeply marked social sciences. In that particular context, Jean Porte forced himself to do some radically empirical work, and tried to group together occupations “which resembled each other” — and that produced the first classification of what he himself called the “socio-professional categories”. Among the essential features of this classification, there was the idea that it was deeply rooted in the conventional structures of the time, even if that had not been fully deliberate. These structures at the time were already completely “naturalized”; they already belonged to the ordinary ways of talking about work and occupations. Another originality of the French classification was that it merged classifications of professions and of social groups. Another important point was that Porte’s classification had two levels: a one-digit aggregate level, with less than 10 categories; a two-digit detailed level, with about thirty categories.

The next step occurred in the late 1970s when it was necessary to revise the classification. This redesign was based on a much more accurate coding system. The third step was the development, more recently, of a European classification, less “idiosyncratic” than France’s (which was so deeply rooted in the specificity of the French context that it could not be used for other countries) and based on criteria sufficiently universal to be generalized.

For Alain Chenu, the only way to understand these changes is to identify which institutions use or have used these representations of society and their associated classifications. According to him, from this perspective, there are two major periods. The first one extends from the Popular Front to 1937, and was marked by the generalization of collective labour agreements and the gradual extension of the so-called “Parodi-Croizat” classification to different branches of society. It is no coincidence that in the first classification, there were miners, distinguished by the fact that they had a collective labour agreement and a specific pension fund. Artisans and merchants, who did not want to get involved in the redistribution pension funds, are also distinguished by the classification, and this can be no coincidence either. The structuring of a classification is never foreign to the national singularities of the history of social protection. The second period, since the 1970s, was marked by the construction of another type of social protection, largely based on the definition of welfare benefits, and which does not – or at least less – refer to the specification of professional sectors. Institutions moved away from what was originally the heart of the classification of social-professional categories. Moreover, the advent of mass education also helped undermine the institutional segregation described, for example, by Christian Baudelot and Roger Establet in their 1971 L’école capitaliste en France (“The Capitalist School in France”). This complicated and “denaturalized” even further the representation of social belonging or membership according to the traditional classification schemes. This movement also went hand in hand with the emergence of the distinction between immigrants and citizens in international agendas. The statutory situations become more and more complex, even within the working classes, as in the case of bus drivers analyzed by Olivier Schwartz as a part of a very large-scale study of which a recent article in Sociologie gives an initial overview (Schwartz, 2011).

Based on his experience both as a trainer for INSEE coding operators, and as an analyst using occupational classifications, Thomas Amossé addressed some of the issues raised in the presentations preceding his own. First of all, he pointed out that these classifications were meant to represent society in the form of social groups, but without relying on a theoretical basis. The theoretical – “constructivist” – position came about later with reforming work led by Luc Boltanski, but the classification still largely remains today an “empirical” tool. Amossé went on to examine claims that the French occupational classification is getting increasingly obsolete – unable to faithfully depict the structuring principles of contemporary French society – and analytically non-homogeneous: it applies social class theory without saying so, and “class” would therefore be, from an econometric perspective, a “bad variable” used instead of another one that remains hidden. Or on the contrary, it would be insufficiently theoretical and thus fail to provide a representation in terms of social classes... Yet, Amossé eventually observed, even if its uses are actually declining, and even though occupational classifications are less and less employed in social science research, empirical evidence nonetheless shows that these classifications are not less relevant than previously, since the inequalities they allow us to identify have not diminished but, on the contrary, have increased. Thomas Amossé concluded by stating that the decrease in the use of occupational classifications is much less pronounced than the erosion throughout the entire population of the global sense of belonging to a social class.

Cécile Brousse headed the INSEE department of occupational classifications during the discussion on the construction of a European classification. She began her presentation by addressing the criticism levelled against occupational classifications in the previous presentations. According to her, one should first welcome the determination to achieve the construction of such a European classification. That said, she stressed how its development is clearly taking place in an unfavourable context, where statistical tradition almost exclusively consists of comparisons between countries, and not between occupational or social groups. To understand how social statistics are treated at the European level, one has to imagine what would have happen if the whole of French socio-economic statistics had been limited to comparisons between regions. So it was a question in which Eurostat was not strongly involved, and the argument used was that professions were not sufficiently established and harmonized criteria for application across Europe. And indeed, many countries did not even have a socio-economic classification of occupations that was stabile. In this context, the credit goes to researchers following Goldthorpe who have tried to develop a European occupational classification. They met in 2004 in what was called the ESEC consortium, and submitted their report in 2007. The classification they proposed was based on the concept of “employment relationship”, measuring the degree of autonomy of the employee, between the two extremes of “labour contract”, strictly subordinating the employee to his employer, and “service relationship” marked by a high degree of autonomy. The categories were thus ordered in nine classes on this scale. Cécile Brousse clearly recognized that there are commonalities between ESEC and the PCS: artisans and independent workers are isolated and separated in both classifications. But she then noted that there are also numerous differences. We cannot mention all of them here but, for example, heads of businesses with more than 10 employees are mixed with managers (“cadres” in French) by the ESEC, which is surprising in terms of the French occupational classification.

Anyway, after 2007, EEC national statistical institutions had to evaluate this ESEC classification, and criticism was strong. For example, the ESEC seemed ill suited to countries of southern Europe where the criterion of the employment relationship would be less important than the distinction between employees and independent workers. Or again, the project was supposed to offer a universal grid, but actually it seemed very “idiosyncratic”… to British society. For other researchers, it failed to empirically validate the theory of the employment relationship, partly because employment relations are not stable over time, especially in the category of “supervisors” – as defined by the ESEC – who are subject to very high mobility, which is problematic for considering them as a category per se. Criticism levelled by the DARES (the research department of the French ministry of labour) and the INSEE also had to do with the absence of considerations of cultural practices and consumption patterns and, therefore, as shown by Thibault de Saint-Pol (Brousse et al., 2010), the ESEC classification is hardly heuristic for analyzing cultural practices. Finally, the INSEE investigated the public ordinary reception of the classification. A survey of 4,000 people showed that 17 percent of respondents failed to identify with a category. Another survey, using card games, showed that it was the executives who managed best to recognize themselves in the representation of the occupational world vehicled by the ESEC classification (for a certain number of such criticism, see Brousse et al., 2010).

Etienne Pénissat, following the presentation by Cécile Brousse, presented the results of a survey on the construction and reception of the EU classification, funded by the French National Research Agency (ANR). He showed that the conditions of production of this classification, and the debate it generated, remained confined to the small world of socio-statisticians, even confined to discussions between INSEE administrators and goldthorpian sociologists.

The Disappearance of Social Classes

For Luc Boltanski, whose presentation ended the morning session, the issue truly lies in the paradox between the maintenance of inequalities, and the disappearance of representations of society in terms of social classes. Basically, the question is: can we separate the object from the ways we observe it, or is it necessary instead to focus on the relationship between social reality and the categories of the observer in a constructivist perspective? We certainly have here, rather than a social construction of reality, an administrative construction of reality, since the role of states seems fundamental to Boltanski in this case. According to him, we currently witness a very interesting new development, of which “benchmarking” is an obvious indicator: ruling institutions, which have considered constructivism as an expression of leftism, eventually discovered that it was appropriate and that they actually could use these tools of knowledge to modify the construction in the direction they wished. Luc Boltanski reminded us that in Le nouvel esprit du capitalisme (“The New Spirit of Capitalism”) he and Eva Chiapello (1999) had tried to show how a whole collection of deconstruction movements had joined together against labour conventions, trade unions and institutions, and how that had been the beginning of the rise of individualism.

The problem is that sociological and statistical tools are partly autonomous, since they are subject to constraints related to their review by peers, by the scientific community, which exercises a form of control. Sociology is thus partly “autonomous” and social classes, as a grid for analyzing social reality, showed more resistance in France than in other countries. Social classes are “sociological” entities, and sociological work on such entities has consisted of attributing properties, or even intentions, to these entities, as in statements like: “workers tolerate less and less chain production rhythms”, which helped delete the individual agent from the description of social mechanisms.

Continuing on from Luc Boltanski’s analysis, François Dubet wondered whether or not this growing lack of interest in social classes, about which presenters have been complaining since the beginning of the day, may be the mirror effect of an exaggerated interest during previous decades, where social classes could be evoked in all situations. The strength of the notion of social class was that it allowed emphasizing the systemic dimension of social reality. And then quite suddenly, that intellectual universe disappeared. But the question can be asked differently: why did we previously see only classes, and not the principles of differentiation that triumph only now in the interpretation of the social world, but were already at work then, such as: age, generation, gender, ethnicity? There is no doubt that simultaneously the origins of social problems have also changed: they are now in the suburbs and no longer in the working class. And concerning the latter, the demand for social justice no longer targets the reduction of inequalities of positions, but equal chances to reach these unequal positions. We no longer want workers to have decent working conditions and access to training comparable to those of managers, instead we want children of workers to have as much a chance as the children of managers to reach a position of manager from which they can continue to exploit workers. So we are in a system where inequalities, violence and exploitation persist, but where this is not reflected in the principles of identification, at least not any longer where we once supposed or expected it before. Dubet concluded that previously if we wanted to convince ourselves that there were social classes, we would go to Flins, were Renault cars were chain produced; today we go to Neuilly, the most well off suburb of Paris.

Stephane Beaud did not disapprove of this last remark by François Dubet. In a speech which was a continuation of a 2007 article on social classes in the journal Movement (Beaud, 2007), he proposed to start once again with the paradox already mentioned by previous speakers – between inequalities and persistent situations of conflict, on one hand, and increasing loss of interest for sociology classes, on the other. Sociologists’ interest and loss of interest do have effects on social reality and its ordinary representations. For instance, it is not entirely absurd to think that the works of Michel Pinçon and Monique Pinçon-Charlot, from Dans les beaux quartiers (“The Nice Parts of Town”, 1989) to Le président des riches (“The President of the Rich”, 2010), partly contributed to spreading the idea that there was indeed a ruling class in France, distinct from other classes and conscious of its interests, and that its epicentre could well be situated in Neuilly. That said, to understand this lack of interest, it may be first useful to take the training of sociologists into account: social classes have gradually disappeared from the curriculums of sociology courses, where students at the same time were moving away from this type of concern, possibly also because such concerns raised issues too close to their own social origins, which they did not want to recall. Second, there are also important methodological issues, for instance in the relationship between sociology and statistics. For statisticians, PCS were no longer reliable, were costly for institutions because they require a lot of recoding, and thus were no longer useful. This widely spread attitude significantly contributed to making debates between sociologists, economists and statisticians more difficult. Yet, those are crucial questions: how should we statistically picture the social world from now on? Eventually, there was a third factor: the injunction of “intersectionality”; that is to say, to mobilize various combinatorial and multi-dimensional categorizations, such as class, gender and race, but in which social classes are actually rather poorly dealt with.

Are Social Networks to Blame?

Luc Boltanski had concluded the morning by renewing a critique already clearly stated in Le nouvel esprit du capitalisme: the development of approaches in terms of social networks may be responsible for the decline of approaches in terms of social classes. According to him, a close examination of the theoretical foundations of social network analysis, as developed after the 1970s, and especially of the two famous articles by White, Boorman and Breiger (1976) and Boorman and White (1976), demonstrates that this approach was explicitly meant to rival analyzes in terms of categories and classes. The presentations of the second part during the afternoon, while within the continuity of the developments of the morning, focused, for some of them, on the possible role of social network analysis in the marginalization of approaches in terms of social classes.

Olivier Godechot began his presentation by noting that the topic “Networks and Social Classes” sounds like a dissertation topic, which, by the way, had caused some controversy last year when “social networks” were included into programs of secondary education in economics and social sciences (Mercklé, 2011). The controversy then suggested that social networks support an irenic representation of social reality, with neither classes, nor conflictive social relations. But to do so is to reduce social networks to networks of sociability, in a rather old fashioned manner. Actually, in the seminal contributions to network analysis, there has always been a very strong relationship between class issues and network issues. Among economists, from the outset, as with Quesnay for example, the fragmentation of societies into different social classes is based on a specific conception of a relational system, namely that emerging from market exchanges. One can say exactly the same concerning the Marxist approach to social classes, which is clearly a relational approach. For another example, in structural anthropology, principles of social differentiation are clearly built on relational systems. So were all the useful elements in the social sciences built with theoretical relationships between classification systems and systems of relations.

One might therefore have expected network analysis to provide grist for the mill of a class-based approach to social structures. On the contrary, the opposite happened, and it was observable from the beginning with the famous inaugural work of John Barnes (1954). Starting with a quest to describe the “social class system” of the village he was observing, the anthropologist claimed he found networks instead. From there on, some researchers turned social networks into a war machine against social classes, as was done, for example, by Maurizio Gribaudi in his “Exercises with Social Networks” (1998), as was done indeed by Harrison White in the works mentioned previously in the morning by Boltanski, or as was done by Padgett and Ansell in the no less famous article proposing a reticular explanation for the rise to power of the Medici (1993). That said, from the 1990s on, we have witnessed the emergence of new forms of possible articulation between classes and networks. From where does a position in the network come? How is the network formed? What is homophily, and how can it be explained? These questions have reintroduced wolves among the sheep, attributes into relationships, social classes into social networks.

After Olivier Godechot’s presentation, Emmanuel Lazega chose to follow through with remarks examining what could be a “neo-structural” position, which would seek to build a model of interpretation of relations between the micro, meso and macro levels of sociology. The analysis of relations between actors enhances the possibility of holding interdependence and conflict together, and to do so in a fairly complex manner. And according to him, there is no doubt that this approach has an untapped critical potential. Network analysis can distinguish the relational resources of individuals and the collective dimension of this “social capital” of relations: forms of solidarity, but also forms of exploitation, of control and of exclusion. We can therefore use these methods to analyze the forms of domination, social control, etc. Dominance is clearly a relational form; it is fully amenable to analysis in terms of networks, which has heuristics virtues, including a critical perspective, and therefore, according to Lazega, one can no longer be content with the criticism by Boltanski of the seminal works of White. Since their publication, there has been an elapse of time of more than three decades, during which network analysis has developed tools to think about it in complex ways – for example, with the work on transformations of networks by Tom Snijders. In his conclusion, Emmanuel Lazega was thus led to suggest that interactions between “networks” (he prefers the term systems of interdependence) and social classes may be appropriately seized using the concept of “opportunity hoarding” which was developed by Charles Tilly (1998, 2005) during the last years of his life, and indeed in contact with Harrison White at Columbia. At the macro-sociological level, society is an organizational form in which one can “cope” by organizing collectively to reach or maintain positions from which it is possible to hoard opportunities: these opportunities do not mainly consist in direct resources, but in chances to access resources without too much difficulty, when they are needed, which can only be achieved by oneself. This includes jobs, finances, apartments, childcare, etc. All these things are accessible in a collective but informal manner, and network analysis is capable, methodologically, of apprehending this level of articulation between the different layers of social processes.

There were also presentations by Frédéric Lebaron and André Orléan. The latter noted that the question of social class was evacuated by a majority of economists on the pretext that we live in market economies, and that the commercial relationship is a universal form indifferent to such specifications. This thesis plays a structuring role in economic thought. However, for André Orléan, this reduction of capitalism to the “market economy” does not hold up, because it excludes from consideration the employer-employee relationship. And what we see is a concomitant eviction of the wage relationship from representations of capitalist economies, and a break in the historical mode of distribution of gains in productivity. Although for many decades, these gains benefited from the increased standard of living of employees, the current evolution is in the other direction.