1How can it balance its budget and even make a “profit” from its services? This is one of the major preoccupations of the Société Nationale des Chemins de Fer Français (SNCF) (the French National Railway Company). In terms of expenses, the company incurs costs in order to run trains on the rail network. In terms of revenue, it receives this from the sale of tickets. The means of fixing transport tariffs is therefore an essential parameter for the financial management of the SNCF.  The issue of tariff setting is not a new one: since the beginning of the railways in the middle of the nineteenth century, the question of the price of train tickets has been one of the major themes in the railway economy (Grall 2003). The solutions offered have nevertheless varied from one era to another. This article offers a historical sociology of rail tariff setting for passengers in France since 1938, the date when the SNCF was created by the nationalization of private rail companies that had previously been using the national network. 
2Highlighting that tariff policies have altered over the course of the history of the French railway sector should offer no surprises if one bears in mind the transformations in the French economy since the interwar years. Three major periods become apparent from an analysis of the changes in the sector (Finez 2013a), which coincide with periods identified by studies of capitalism and the state in France.  From its creation, although the SNCF in part inherited the commercial aspirations of private companies (Caron 2005: 295–324), it was above all “technicist” and “productivist” considerations that prevailed in the railways during the decades that followed nationalization in 1938. Like other large companies nationalized following the Liberation in 1944, for example Électricité de France (EDF), Gaz de France (GDF), Charbonnages de France (French national electricity, gas and coal companies respectively), the SNCF was initially entirely devoted to the reconstruction of the national economy and the modernization of the country (Kuisel 1981). Then, from the end of the 1960s, the SNCF management model seemed to converge towards to that of private sector companies: the public rail operator made increased use of sub-contractors, opened a marketing department, was reorganized into commercial branches, and used econometric tools to simulate demand. Some authors saw these changes as the first forms of “privatization of the sector” (Dubois 1974). From the 1990s, commercial changes to the SNCF accelerated. The “Groupe SNCF,” now made up of several hundred subsidiaries, expanded internationally and diversified its activities. This second liberalization is one of the components of the European “neoliberal turn” (Jobert 1994) which, along with railways, saw a number of public sector utilities considered “natural monopolies”  switch to the free market: for example, electricity, gas, telecommunications, postal services and air transport.
3However, the aim of this work is not so much to illustrate one of the prominent aspects of the intensification of market regulation in the railways but more to provide an analysis of the production conditions of tariff systems specific to the SNCF. This article—which is in the tradition of other sociological studies on prices, notably those by Bidet (2010), Barrey (2006) and Yakubovich et al. (2005)—is interested in the tools, actors and contexts that contributed to the genesis of tariff discrimination in the railways and in doing so contributed to redefining the eminently polymorphous principle of public service in the rail sector. Tariff discrimination can be defined as the differentiation in the price of a journey in terms of the characteristics of demand or costs. Therefore, price setting is not the result of a “natural” matching of supply and demand, but of “tariff setting” (Barrey 2006) carried out by a few identifiable actors and whose concrete activity I attempt to describe. These “entrepreneurs in economicity” (Steiner 2005)  use economic science senso lato along with their influence to attempt to impose what they consider to be a “good price,” a fair price. Far from operating in a structural vacuum, these SNCF price setters act within specific (political, social, technical and competitive) contexts that constrain their actions but also render them possible, opportune and sometimes necessary.
4The story of tariff discrimination in the French railways unfurls in three stages. In 1938, when the SNCF was created, prices were based on a uniform per-kilometre tariff for the whole rail network. During the post-war years, this tariff equalization system was criticized in favour of a second tariff paradigm. Based on the tentative integration of the costs involved in the price of a ticket, marginal-cost tariff setting was adopted in the 1960s–70s. Soon considered ill suited, it was replaced by a completely different mechanism. At the turn of the 1980s–90s the SNCF put in place a system of real-time differentiated tariffs called yield management. The objective of this mechanism was to fix prices in terms of the passengers’ propensity to pay to travel on the network. In a more complex form, this is the tariff model that still prevails in France today for travel on the “grandes lignes” (mainlines). From 1938 until today, three tariff structures have thus succeeded each other in the SNCF. They have been accompanied by a transformation in the nature of the service sold and the characteristics of the passengers transported. Initially uniform and presented to an undifferentiated clientele, transport services have gradually been broken up to be sold to groups of targeted clients in order to balance the company’s books, then to consumers with heterogeneous preferences with the aim of capturing the “social surplus” for commercial gain.
5The article is in four parts. First a broad overview picture of the existing sociological literature on prices is painted in order to adopt a theoretical position and to highlight the originality of this research. The three following sections are devoted to an empirical analysis of the changes in tariffs at the SNCF: how the tariff equalization system that initially prevailed in the French railways worked, an analysis of the move to marginal-cost tariff setting and the adoption of yield management.
Box 1. Methodology
I have systematically studied the Revue Générale des Chemins de Fer (RGCF) [†] from 1938 to 2012. I also used a series of administrative reports devoted to public companies in general and rail transport in particular, grey literature (memoranda, research reports and theses concerning the SNCF), personal accounts, essays and scientific articles written by the major actors in the sector as well as newspaper articles. Moreover, I was able to socially position the prominent actors and trace their career paths by consulting biographical dictionaries such as Who’s Who in France and directories of former students of the grandes écoles. 
Finally, I draw on 11 interviews as well as exchanges of correspondence with the main instigators of tariff models implemented over the last thirty years (SNCF managers, senior directors, managing directors) and with trade unionists, mostly opposed to the changes described.
The theoretical framework
6How are prices set? What are the resources, mechanisms and tools used by actors to define the sum of money for which a good or service is exchanged? What are the social institutions that constrain and direct this particular type of exchange? The last fifteen years has seen an increase in the number of sociological studies on price setting, in France in particular.  These studies often analyse financial markets (MacKenzie and Millo 2003; Muniesa 2003). They are also concerned with “markets of singularities” (Karpik 2007) where products are non-standardized and of uncertain quality, in particular the wine market (Benjamin and Podolny 1999; Chiffoleau and Laporte 2004) and the art market (Velthuis 2005; Beckert and Rössel 2004). Studies on prices in mass markets have received less attention.  There is, however, increasing work on the setting of prices for standardized goods and services: for example in retailing (Barrey 2006), meat marketing (Anzalone 2009), electricity distribution (Yakubovich et al. 2005; Poupeau 2007) and telecommunications (Bidet 2010). Our hope is that this research on rail transport price setting will contribute to the debate.
7Two types of criticism are sometimes made of the sociology of prices: it seems important to us to outline these and take a position in respect to them. The first criticism concerns the heterogeneous character of the work on this subject. For Beckert, in a recent article on the state of the art on this question, sociologists agree that prices are determined by “social and political forces operating in market fields as well as the social and cultural contexts forming the preferences of actors” (Beckert 2011: 757). But beyond this a minima point of convergence, with the exception of Harrison C. White’s (2002) theory of markets, there is no synthetic theory of prices that dominates in the same way as there is in economics. Ultimately, this is not surprising given the diversity of markets studied by sociologists: creating a unified theory would efface the plurality of social mechanisms at play in price setting. But should we not adopt an approach similar to that of Geertz in his study of the Sefrou souk? Far from proposing a general model of practices to replace standard economic models, the anthropologist chose to construct a specific economic type—what he calls the “bazaar economy”—which enabled him to closely study his subject, but could also be used by extension as an interpretive framework for studying other markets (Geertz 1979).
8The second criticism concerns the ambiguous relationship between the sociology of prices and economic theory. In an article on the contemporary economic sociology of prices Fillieule asserts that these studies are always, but most often implicitly, based on economic models: in general the “law” of supply and demand, the auction model, or the “law” of costs (Fillieule 2008). There are few sociologists therefore who adopt a position in relation to economic theory.  Perhaps this should be seen as a kind of inductive reflex of the researchers in the discipline, who consider that economic “laws” “do not generally merit this label” and in reality are most often “maxims for action, or in reality practical precepts” (Durkheim  1982: 68). Sociologists are no doubt also aware that many commercial exchanges that at first sight seem to respond to the application of common economic principles are in reality often based on other mechanisms: such as judgement devices, information flows determined by market structures, the social relations in which they are rooted, norms produced within or outside of commercial institutions, etc. In other words, most of the time, prices are not set in an endogenous way within a closed and restricted world of a neoclassical market. The market should therefore be “re-peopled” and “re-institutionalized,” just as this article proposes. This analysis of the process of constructing tariff models in the SNCF therefore describes how certain formulas for calculating prices have been established.  It also shows how these price formulas are always linked to certain moral requirements, such as the principle of general interest, which in the past as today characterize the rail sector but whose definition has varied depending on the era.
The study of SNCF prices as an analysis of a state-controlled tariff
9Justifying the use of the term “tariff” as meaning the sum necessary to buy a train ticket enables us to specify our approach to the subject. Let us first recall that the transactions studied concern relatively homogenous goods whose prices—more precisely price systems—are durable over time and in so being can be publicly displayed. The etymology of the word “tariff” is a reminder of this.  It should also be pointed out, as Steiner highlights in his economic sociology book on organ transplants, setting a tariff is the result of work by experts who devise models conforming to the requirements of governments and not the result of a simple meeting of commercial interests as is the case for market price setting (Steiner 2010: 204–5). The work of tariff construction (rail, postal, electricity, etc.) results in the writing of a script, to use an expression from the anthropology of technology. This script, which can be revised from one period to another, in this case is a mathematical formula or algorithm that enables the price of each train ticket to be determined.
10In a public company such as the SNCF, the state (through its representatives) is involved in the rules for setting prices for the goods exchanged. Does this mean, however, that the public authorities do not intervene in price setting in other markets? Answering this question in the affirmative would be to deny the role of the state in the economy, a field in which it is probably most involved (Bourdieu 2000). A tariff, even when not set by the state, is regulated by the legal framework that restricts the conditions for exchange and influences price setting more or less directly. Safety regulations (health, food, etc.) relating to the selling of consumer goods, labour legislation and more generally all the regulations that govern commercial exchanges (Fligstein 1996) are constraints that guide companies’ pricing policies. But the public authorities are still more influential in the French railways and, more generally, in all historically monopolistic utility sectors (electricity, gas, coal, postal services and telecommunications) than in other markets.
11The state is both the owner of the SNCF and financial guarantor of the company. Thus, in contrast with the private sector where a firm’s repeated losses would mean its eventual dissolution, market sanctions are only indirectly applicable to the French railways. Two types of situation are possible. If the state owner decrees that the SNCF’s priority is not to make a profit or to balance its books, it may well decide, if necessary, to replenish the company’s coffers each year from public funds. This is the system that tended to prevail in France until the end of the 1950s. A contrario, if the state thinks that the SNCF should meet its public service obligations—the definition of which varies from one era to another—by limiting its recourse to public funds, self-financing and even making a profit, it can force it to reform its management and in particular to revise its tariff setting system.
The contexts and power relations underlying the “work of tariff setting”
12Given that the price of a train ticket is a tariff, it does not materialize ex nihilo. We must, therefore, look at the work of “price setters” who reject certain tariff systems and construct new, more appropriate ones that conform to the principles of fairness they support, but also identify who the tariff setters are with a view to understanding their motivations. As such, if we compare the social profiles of price setters to those of other agents in the SNCF, their uniqueness is clear: the engineers, educated in mathematical economics in the 1950s–60s and the young graduates from business schools of whom some came from the private sector in the 1980s had social characteristics that dispose them to act against the prevailing doxa in the rail “field” (Bourdieu 1980a) and in so doing to innovate. Price setters rely on economics in the broad sense (Muniesa and Callon 2009) to transform tariff systems. Economic theory, statistics, econometric forecasting techniques, market research tools and also current practices observed in other economic sectors—in particular air transport in the United States—are all resources they use to justify the adoption of new price paradigms.  The influence of economics on social reality should also be considered in light of the interplay of actors and interpersonal networks that enable the diffusion of marginalist thought and the gradual evolution of price formulae in the railways: the rail tariff setters thus use the ties they have with senior civil servants and SNCF directors, who are often former classmates at the École Polytechnique or were teachers at the grandes écoles of which the price setters are alumni. 
13Accurately analysing the process of SNCF price model transformations requires, moreover, placing it back in the context of the politico-institutional environment that surrounds it (Fligstein 1996). Since the early years of the Fifth Republic, an increase in respect for technico-economic expertise has been observable. Economists close to power and senior civil servants seeking to rationalize the state, in particular within the Ministère des Transports (Ministry of Transport) (Mazoyer 2012), contributed to the promotion of economic expertise (Dulong 1996).  The actions of this modernizing “small group of pioneers” (Gaïti 2002) benefitted from the development of planning in France (Fourcade 2009). It was also nourished by the socialization to economics of future elites encouraged by the introduction of the discipline to the curriculum of the École Nationale d’Administration (ENA) (National School of Administration) and the Institut d’Études Politiques (Institute of Political Studies) in Paris (Kolopp 2011). At the turn of the 1980s–90s, “economic belief” (Lebaron 2000) further extended its sphere of influence. The development of new public management (Bezes 2009), and the spreading of a “market culture” from the United States and the UK within certain grandes écoles and business schools where the public and private elites were educated (Lebaron 2009) is one of the expressions of this. It is thus also the legitimization of recourse to economics and to its theories and models within the upper echelons of the state that facilitated tariff innovation in the SNCF.
14Setting a price is also “carried out with” the tools and techniques available to the tariff setter as well as the competitive context (Barrey 2006), all are elements that define the contours of the space for possible tariff formulae. The challenge to rail transport’s de facto monopoly in terms of transport (competition from road and later air transport) and anticipation of the legal opening up of the rail sector to competition have tended to be presented as requiring tariff innovation. The development of information technology and marketing techniques  made the development of yield management possible during the 1980s–90s.
15Finally, the work of tariff setting is tied up with conflicts and the evolution of power relations between the actors (individuals and groups) involved in the French railways. The “Wright system” model for electricity pricing adopted in the United States at the end of the nineteenth century was not the most efficient but was the most favoured by powerful industrialists in the sector (Yakubovich et al. 2005). The company strategies (of which price is one of the elements) that were adopted in the same era in the US rail sector were the result of a particular interpretation of the legal framework (US antitrust law) imposed by economic actors in positions of power (Dobbin and Dowd 2000). In our case, it was the overhaul of the SNCF’s senior management in favour of directors who supported price discrimination, along with a lack of interest in tariff issues from trade unions that enabled tariff innovation.
Tariff equalization as the historic tariff setting model
16Our study begins in 1938, when the SNCF was created by the nationalization of private rail companies. The change in economic model reaffirmed the tariff policy that had prevailed until then. Tariff setting for passenger rail transportation in France was based on the principle of tariff equalization.  This meant that all passengers were charged a ticket price dependent on the length of their journey, whatever company or line used by the traveller. Tariff equalization should be understood in light of the strategic role of the railways for the nation. In addition to military concerns, the railway network was a tool enabling the unification of the country and an increase in its wealth by conveying goods throughout the land and encouraging citizens to travel. “The transport tariff is the great lever for public wealth” (Marqfoy 1863: 5), noted a railways engineer in the mid-nineteenth century. Tariff equalization was also an expression of the principle of equality of treatment, which was one of the two great pillars of the SNCF’s cahier des charges (document setting out the SNCF’s obligations), the second being the obligation to transport. Note, however, that no legal regulation obliged the public rail operator to adopt a uniform per-kilometre tariff. The cahier des charges even specified that tariffs could include measures to ensure the staggering of traffic during peak periods.  But such measures were practically nonexistent since they were unpopular within the company and the government as well as among passengers (Bourgeois 1946: 343).
17The question of tariff equalization is generally examined as a whole, but it could nevertheless be considered to have two distinct forms. Geographic or territorial tariff equalization is based on the fixing of a single per-kilometre tariff that is valid across the whole network. The financial management of the network is based on the principle of cross-subsidization: profitable lines and sections finance those that are less profitable or loss making. Although less commonly used, the principle of temporal tariff equalization is based on the existence of a single price for travelling on the same line whatever time of day or day of the year a train is running. Profitable journeys finance others in the same way.
18The cost of running a train whatever line, section of line or time slot depends on many variables, in particular the number of passengers on board. In contrast to goods transport, which is less affected, irregular traveller numbers characterize passenger traffic. This phenomenon results in large regional, timetable and seasonal variations, which tend to combine. On certain lines there are spikes in usage at the beginning and the end of the day as well as at the beginning and the end of the week as a result of journeys between work and home. Even more acute peaks occur at certain times of the year, in particular during certain religious holidays and in the summer holidays. A contrario, in the middle of the day and at the end of the afternoon, in the middle of the week and outside of the holiday season many trains are under-occupied thus resulting in off-peak times. The difference in occupancy and the number of trains running in “off-peak” and “peak” periods can be very large (Figure 1). In 1970, the ratio of the number of passengers travelling on the highest peak day (Saturday, 1 August) and the day when occupancy was at its lowest was 4.6:1 at the national level.  This phenomenon, moreover, has tended to grow over the decades, in particular as a result of the increasing mass popularity of holiday departures following the introduction of paid holidays in 1936.
Peak and off-peak periods, 1938–39
Peak and off-peak periods, 1938–39Note: Significant seasonal variation can be observed in the number of trains departing from and arriving at the Paris-Lyon train station. Peak occupancy (summer holidays, Pentecost, Christmas holidays, Easter holidays, etc.) and off-peak periods (October–November, January–February–March, etc.) are particularly costly for the SNCF.
19Other than social tariffs financed by the state and aimed at specific categories of passengers (children, large families, military personnel, etc.) and a few other commercial tariffs offered by the company itself (group tariffs, return tickets, weekend tickets and annual holiday tickets), the price of train tickets was set based on a uniform per-kilometre tariff. For example, in 1944, the price per kilometre ranged from 1.19 francs in first class to 0.84 francs in second class. To work out what a passenger should pay, all that had to be done was to consult the “indicateur Chaix”  and take note of the length of the passenger’s journey. The price formula is thus:
20Px→y = Dx→y × C
21where Px→y is the price to be paid by a passenger departing from station x and wishing to arrive at station y, Dx→y is the distance between stations x and y, and C is a constant corresponding to the price per kilometre (in second class, for example). 
22Despite irreducible fixed costs (associated with issuing train tickets, station costs, ticket collection, etc.) there was no supplementary charge and no principle of “degression” (reduced per-kilometre tariffs) on long-distance journeys. Similarly, although increasing the speed of trains involves additional costs, there was no surcharge on speed (Bourgeois 1946: 357–65). The passenger transport service could therefore be considered to be a relatively uniform product, serving a clientele whose preferences were undifferentiated.  An important point should however be made so as not to naturalize the prevailing model of cost effectiveness at the time. The per-kilometre tariff did not exist outside of any metrological framework: the distribution of costs between the different train tickets sold involved an economic calculation, as indicated by the “passenger-kilometre” principle, the unit of measurement used by the SNCF to quantify the volume of demand. In contrast to other tariff models based on the principle of flat-rate prices whatever a journey’s length (national postage of letters and parcels, public transport in many French towns, etc.), in the railways the price to pay varied depending on distance travelled. Tariff equalization was based on a classic principle of commercial fairness: for a given service, defined here by the distance travelled, it was considered fair that each person should pay the same price. This conception of tariff fairness accorded with the impulsive economic morality of the Polytechniciens in the grands corps  (Grall 2003), who had managed the sector since the nineteenth century. The per-kilometre tariff also conformed to the standard economic views of passengers, used to paying depending on distance travelled.
The construction of a new pricing paradigm: Marginal-cost tariff setting
23How might irregularities in demand for transport be dealt with and how could the company’s books be balanced? This was one of the major questions for the SNCF in post-war France. If a government allowed it to, a public company in a monopoly position could reduce its losses by raising prices. But when its clientele is no longer a captive one, when competition between modes of transport develops, financial difficulties will accrue. Until the inter-war years the train was the principal means of transport because of its speed and affordable prices. The rising number of motor vehicles and roads nevertheless changed the game. Competition between modes of transport was initially limited to goods, in particular along the main routes. The inflexibility of the railways in terms of schedules, routes and tariffs prevented the SNCF from reacting when lorries captured the most profitable markets in the 1940s. Moreover, from the end of the 1950s, the number of private vehicles rose sharply, breaking the ten million barrier in 1966 (Neiertz 1999: 395). More flexible in terms of schedules, more comfortable in terms of journeys, the car benefitted from new and modern infrastructures, such as motorways. Passenger rail transport, which had up to this point profited from a captive clientele, also found itself facing competition.
24In the eyes of a section of senior civil servants in the ministry in charge, this situation was untenable. The development of the railways, an activity benefitting the nation as a whole, should certainly be promoted, but a solution had to be found to the problem of financing the SNCF, a company that had been making permanent losses during the post-war period despite large contributions from the state.  The engineers from the technical corps who staffed the Ministère des Transports and the management of the SNCF were opposed to challenging the historic tariff-setting model, in particular with regard to passenger trains. Some engineer-economists,  influenced by Maurice Allais’s theory of social efficiency, thought in contrast that the proper management of natural monopolies would be done through tariff de-equalization.  They asserted, supported by mathematical demonstrations, that abandoning the per-kilometre standard would enable the company to balance its books while also promoting the general interest. Economic theory had a remedy: marginal-cost tariff setting. These tariff setting changes, initially applied to the transportation of goods, were the backbone of reforms in the transport of passengers during the 1970s.
The tariff setting work of Allaisian engineer-economists
25In a basic system like tariff equalization, accounting for costs is relatively easy. Simply define or reassess the price of a standard kilometre based on forecasting operating expenses, the number of passenger-kilometres and tonne-kilometres for the coming year. This simple technique, nevertheless, becomes a constraint when one wishes to differentiate transport costs. All the more so given that simply determining operating expenses on the railways proves to be a “formidable undertaking” (Hutter 1950: 63). Changing tariff setting requires the construction of a new framework for measurement that meets specific management objectives—pursuit of a balanced budget—, and justifying that it conforms to the principles of public service set out in the Cahier des Charges de la SNCF.
26From the 1940s, a theoretico-empirical rethink on tariff setting in the railways began. It was conducted in 1947 by Roger Hutter and Marcel Boiteux, two economists close to Allais  who endeavoured to show the benefits of constructing a tariff based on marginal cost and not on a journey’s length and the goods’ value.  Marcel Boiteux related his own experience at the SNCF thus: “One day Maurice Allais said to me: ‘My friend Hutter would like someone to help him think over remodelling rail tariffs, you should go and see him and work with him.’ So I started examining the problem. … When I said that tonne-kilometre prices should be different for when travelling full and travelling empty, I was looked at wide-eyed. This was considered totally extraordinary because the tonne-kilometre was a sort of transcendental thing that one was not entitled to question” (quoted by Bungener and Joël 1989: 25–6).
27Applying marginal-cost tariff setting—more precisely at the “marginal cost of development,” in order to anticipate costs associated with traffic peaks over the year to come—means selling a train ticket at the cost of producing an additional unit. Prices thus vary according to journeys, seasons, train time and rate of seat occupancy. According to Allais’s theory of social efficiency—the theory for which he received a Nobel Prize in 1988—, cost transparency and a strict application of marginal-cost tariff setting in rail transport are a sine qua non condition for the maximization of social efficiency, which is synonymous with the general interest (Allais 1948). Reflecting the cost of production in a price signal enables consumption to be rationally directed. Avoiding distorting the choices of network users guards against wasting productive resources, the user adopts the most suitable means of transport (train or car) depending on their needs and the value they attribute to using that service. Thus, tariff setting based on marginal cost, because it can mimic a perfectly competitive market by calculating a virtual price to aim for, is the best management alternative in a natural monopoly like the railways.
28The marginalist precept supported by the engineer-economists aimed to optimize the economic management of the business by estimating the market. It is true that optimization issues were far from alien to the engineers and were a good fit with the “computational thinking” (Vatin 2008) of the Polytechniciens in the grands corps who managed the SNCF at the time.  Nevertheless, during the period of post-war economic reconstruction and growth, the “économie de forces”  dominated. This meant producing more each year by rationalizing use of the network and materials and minimizing the use of fuel by locomotives and maximizing wheel traction on the rails. Another issue was how to respond somehow to the demand for passenger transport by running relatively comfortable trains that ran faster and faster, arrived on time whatever the technical limitations while making it a point of honour to meet the numerous safety regulations. The issue of balancing the company’s books was certainly not completely ignored, but it was not an essential issue for the SNCF management, with the state acting as a guarantor. Like other key industrial sectors, the essential aim of the railways was primarily to contribute to the modernization of the French economy (Kuisel 1981).
29Thus, while Allaisian engineer-economists believed they had theoretically demonstrated that tariff de-equalization was an instrument for making users and public companies accountable and proposed a practical solution to the problem of competition between road and rail, SNCF engineers and senior civil servants continued to assert that public service rail transport required a uniform price for each kilometre (Boiteux 1993: 46). Economics cannot change reality in an autonomous way. To have an impact on the world, a new theory must not simply be stated but must also be supported by influential actors who diffuse it through various channels. At the time, Allais propagated his vision of the economy in the private sector and in public companies, as well as in the upper echelons of the state through his interpersonal connections with senior managers and civil servants and also via his ties with the engineering grandes écoles (Denord 2007; Fourcade 2009). He also spread his ideas through seminars, such as the one he led at the Café Cérou in the Place Saint-Sulpice, Paris, that brought together a dozen people coming, in particular, from the SNCF (Bungener and Joël 1989: 19). At the same time, Hutter recommended marginal-cost tariff setting and the need for profound commercial reform of rail transport in three articles that appeared in the Revue Générale des Chemins de Fer, the reference journal for SNCF engineers (Hutter 1950). In addition to a famous article on this subject that appeared in the prestigious journal Econometrica which led to him being elected president of the Econometric Society (Boiteux 1956), Marcel Boiteux spoke highly of the virtues of marginal-cost tariff setting in public monopolies at the École Nationale des Ponts et Chaussées.  During the post-war years, marginalist thinking therefore spread and tended to permeate and stimulate debate on reforming the management of public companies.
The origins of tariff de-equalization: A crisis of legitimacy for the SNCF management
30The political conditions for a reform in tariff setting emerged at the turn of the 1960s. The Rapport sur les obstacles à l’expansion économique (Report on the obstacles to economic growth) led by Jacques Rueff and Louis Armand  drew attention to the existence of a distortion between costs and prices in rail transport, leading to large losses for the SNCF. This was damaging to the development of the national economy, in particular within the context of the Europeanization of trade following the signature of the Treaty of Rome in 1957. Members of the committee then suggested that for the transport of goods by rail there should be “a moderate and progressive de-equalization of tariffs, resulting in some cases either in these being lowered or raised” (Premier Ministre 1960: 22). A guidance note stated that it would be appropriate to “bring to a conclusion the studies by the SNCF to ‘deequalize’  their tariffs, meaning now setting tariffs dependent on their real costs, varying according to journey, time and goods.”  The committee’s recommendations were heeded. By weighting each goods transport service based on cost, the 1962 reform was the “greatest upheaval to tariffs undertaken since the beginning of the railways.” 
31The Rueff–Armand report only dealt with the rail transport of goods. But such a reform filtered through to promote reform of passenger transport tariffs. Members of the committee thus said that, in general, the SNCF “should determine its commercial policy based on the real costs of the services it provides” (Premier Ministre 1960: 23). Such recommendations were the first evidence of changes to the mindsets of the politico-administrative elites. From the 1960s, the idea that, like the major government departments, the SNCF was a bureaucratic and paralysed organization (Crozier 1970) began gradually to spread. Railway engineers were increasingly criticized for developing the business along productivist lines, without worrying about the effect their behaviour had on public finances. Numerous experts and economists, who were increasing in number and influence within the Ministère des Transports (Mazoyer 2012), along with a growing proportion of senior civil servants in charge of the sector believed that the SNCF should no longer seek to satisfy demand for transport at any cost. It was not the public that should finance the running of the business from the public purse, but customers—that is both “shippers” (goods transport) and passengers—through an appropriate tariff structure.
32A well-known report coordinated by Simon Nora  crystallized this redefinition of the place and role of public companies in France. In addition to the savings that public monopolies should make, the report advised that the state owner should grant them more commercial autonomy: it was in this context that the first programme contract was signed between the state and the SNCF in 1969. According to the Nora report, public companies should also be encouraged to be more “inventive” and “rational,” urged to modernize their management style drawing on the private sector, which was considered to be the model to follow par excellence (Nora 1968). In short, as the Prime Minister, Jacques Chaban-Delmas, suggested (Assemblée Nationale 1969: 2253), it was necessary to “make public companies be true companies by restoring control over decision-making to them.”
New company strategies and new leaders: The beginnings of tariff discrimination for passenger transport
33It took time for the mindsets of the political elites to change, and also to convince the engineers in the grands corps working in the rail-company and ministries of the merits of marginal-cost tariff setting.  Studying the make-up of the SNCF management since its creation shows a change in career profiles of staff beginning at the turn of the 1960s with the arrival of a new generation of engineers—this also reflected changes to ways of thinking within the state.  The advocate of marginal-cost tariff setting, Hutter, who attained the position of deputy director general in 1966—and who supported the new director general, Guibert, a former student of Allais—, also called this period the “colonels’ revolution” (Picard and Beltran 1994: 37). As was customary, the new directors of the SNCF were Polytechniciens from the state’s technical corps (Corps des Ponts et Chaussées and Corps des Mines). But in contrast to those who had managed the company up to then, the new senior managers were interested in economic theory and more specifically Allaisian mathematical economics, and wanted to redirect the management of the railways in line with a vision they called “technico-economic.” After several years of experimentation, it was in 1970 that it was established that ticket tariffs had to better reflect fixed costs, especially on short journeys. A fixed supplementary charge was introduced into the tariff setting formula. The increase resulting from this fixed charge could not however raise train ticket prices by more than 15%.  The formula for this “binomial tariff” is expressed thus:
34Px→y = A + Dx→y × C
35where A is a fixed supplementary charge.
36Although limited and tightly controlled, this early tariff discrimination was a major qualitative and symbolic leap. But for the advocates of marginal-cost tariff setting, it did not go far enough. The new SNCF marketing manager, Jean Ravel, still complained of having to “fight against the inertia inherent to all public services” (quoted by Montbrun-Guttierez 1978: 223). This atypical manager—he was not a Polytechnicien and built a career at Air France—was the first to recruit young graduates from the business grandes écoles on a “large scale,” in particular from the École des Hautes Études Commerciales (HEC) in Paris, by going directly to the schools to find students.  Helped by these new recruits, he wanted to take railway management in a commercial direction, in particular by introducing the principle of timetable and seasonal variation. Based on extensive surveys, the marketing department produced a new profile of typical passengers. SNCF users were not a homogeneous clientele, but customers with diverse needs and expectations. That was why the company had to strive to meet demand for transport in a more appropriate way, in particular demand from the wealthiest passengers (Lacôte 1974: 14). Thanks to the knowledge gained from the market research work, and following on from the change in the grammar of tariff setting, it was possible to increase the SNCF’s revenues and thus to balance its books. To do this, the clients who were most price sensitive or were the most flexible in terms of when they could travel were to be encouraged to journey at off-peak times in order to even out the peaks in demand.
37In September 1979, at the initiative of Ravel who had since become passenger sales manager, the SNCF introduced new deals (Couples, Family, Senior and Leisure tariffs) and a three-colour calendar. Time slots during the year were classified in three colours: blue for “off-peak,” white for “normal” periods and red for peak times. Symbolically the new SNCF tariffs were similar to the “yellow tariff” and “green tariff” introduced for electricity by the EDF in the 1950s. But the SNCF’s three-colour calendar was inspired more by the “blue–white–red” colouring of tariffs introduced by the Canadian National Railway in 1962.  This system enabled the Canadian company to increase traffic by more than 60% in five years and to quickly achieve financial equilibrium (Lacôte 1974: 219). Henceforth, the price of a train ticket would also vary as a function of demand for transport during a time or calendar slot. A parameter was thus added to the price formula, which could now be expressed thus:
38Px→y = A + Dx→y × C × ∆t
39where ∆t is a parameter whose value varies depending on period in the three-coloured calendar.
40It seems that blue–white–red tariff setting was a commercial innovation that was introduced with little fuss. The Confédération Générale du Travail (CGT) (General Confederation of Labour) and Confédération Française Démocratique du Travail (CFDT) (French Democratic Confederation of Labour), which were the two main trade unions at the company, were certainly opposed to the SNCF’s commercial strategies from time to time, but trade union declarations at the time mostly reflected preoccupations concerning downsizing and station and line closures. Any objections to the tariff paradigm have not, in any case, been formative or been a constituent element of the “identité cheminote” (“railway worker’s identity”).  The SNCF did not lack arguments to support the new tariff: thanks to time variation, it better managed the network and limited the company’s losses despite growing competition from road transport, which also enabled it not to increase the average price per kilometre. Moreover, the rail company continued to offer some tickets without additional charges, even reductions on previous tariffs, presenting less wealthy customers, who might not normally have travelled, with the opportunity to take the train. But to do this they had to accept travelling off-peak (“blue” tariff). The new SNCF pricing policy thus highlighted the change of reference for economic fairness based on operating costs, which was established in the railways during the 1960s–70s. But the development of the train à grande vitesse (TGV) (high-speed train) programme presaged a new way of setting tariffs, substituting the measurement of marginal costs by the utility each passenger associated with a transport service.
TGV tariff setting and yield management: A symbolic revolution
41The three-colour calendar enabled the SNCF to generate additional revenue, but the new system was still very crude. Indeed, the coloured periods were fixed annually and applied in a uniform way across the whole network. The inflexibility of such a system did not therefore allow the diversity of situations that might arise, even on the same line, to be taken into account. Trains from Paris to Lyon and from Lyon to Lille would attract the same supplement on a Friday evening while only the former route was likely to be busy. Moreover, fixing the calendar once a year left no room for uncertainties. How could decisive but unpredictable factors like the weather be anticipated? There was therefore a drive to change the price formulae. The details remained to be outlined. The most notable changes were directly linked to the TGV, technology that was developed to compete with increasing competition from air travel. An indication of this is that, from 1969–82, Air Inter’s traffic increased from one to five billion passenger-kilometres (Neiertz 1999: 516), the majority of flights concentrated on a limited number of routes connecting the capital with a few provincial cities like Lyon, Bordeaux, Marseille, Toulouse and Nice.
Abandoning the kilometric scale
42By removing the principle of the kilometric scale, the 1980s reforms marked a significant turning point in the history of railway tariff setting. In 1981 the first chink appeared in a subtle way—because it was invisible to and painless for passengers—with the introduction of the first high-speed train. To attain unparalleled speeds on rail, TGV technology required special rail tracks. The ligne à grande vitesse (LGV) (high-speed line) built between Paris and Lyon was 87 kilometres shorter because it avoided the town of Dijon. However, passengers continued to pay the same price as they would to use the traditional Paris–Dijon–Lyon line. The SNCF wanted to make a quick return on the investment associated with developing the TGV project.  Moreover, at certain times of the day or week, ticket prices were increased by 20%. Finally, just like the airline industry, the SNCF implemented an obligatory reservation system with a charge (+ 8 francs in 1981). The TGV Sud-Est project was intended to win back customers from higher socio-occupational groups. Carrying standing passengers was therefore out of the question. The price formula for a train ticket on the TGV Sud-Est line is expressed thus:
43Px→y = px−y × N + R
44where px−y is the basic price for journey x–y, N is a coefficient equal to 1 in normal periods, and 1.2 in peak periods and R the price for reserving a seat on the train.
45The tariff reform that accompanied the development of the TGV was perceived by railway workers’ unions as an attack on the traditional model of public service.  However, it did not give rise to any specific protests.  Further tariff complication followed over the course of the decade, reinforcing the SNCF’s marketing policy initially developed for the Paris–Lyon TGV. In order to limit competition from air transport over long distances, the SNCF introduced the principle of degression of kilometre prices in 1986. Alongside this the company launched the Joker programme with a view to filling the least occupied trains: prices were no longer linked to distance but fixed according to market conditions.  To qualify for such tariffs, passengers had to buy their tickets in advance (J-8, J-30, [8 or 30 days in advance] etc.). Then, authorized by the new plan contract signed with the state, the SNCF introduced the Résa 300 system when launching the TGV Atlantique in 1989: prices for the high-speed trains would henceforth vary according to four levels. The tariff changes in the 1980s clearly involved a redefinition of rail transport services. While passengers benefited from the greater comfort (guaranteed seats) and speed of the TGV, it would no longer be possible to board trains without having first booked a seat. This change enabled the SNCF to know the occupancy rate of each train in real time and assess customers’ price sensitivity, the data at the heart of the emerging tariff setting system: yield management.
The origins of yield management: The importation and implementation of a new tariff setting model
46At the end of the 1980s, eager to “dust down” an SNCF that was considered aging with respect to the functioning of the French economy, but also convinced of the imminent liberalization of European railways and the expansion of air transport, some executives wanted to reinforce the market orientation of the company. The director of the SNCF passenger branch, Jean Marie-Metzler,  thus decided to set up a new work group: the Groupe de Recherche Long Terme (GLT) (Long-term Research Group). The composition and organization of this group of a dozen people contrasted with the professional culture and esprit de corps of railway workers at the time. The GLT had a flexible and informal structure, which did not feature in the organization chart of the company, although the SNCF was reputed to be very bureaucratic. Also, and unusually, members of the group were mostly recruited from outside the SNCF, a contribution of “fresh blood” that would encourage the emergence of new ideas. GLT managers were young (in their thirties on average), often business school graduates and, in part, from the private sector. Finally, and doubtless most importantly, members of the team had a more competitive vision of the business. It was for this reason that they were recruited.
47The GLT wanted to propose a profound upheaval of the price models in operation at the SNCF. Different options were studied, among which was an intensification of marginal-cost tariff setting, tariff setting “at market prices” and also the implementation of a yield management system. It was the latter solution, considered the most innovative, that would be adopted.  Yield management is a tariff setting scheme that emerged in the highly competitive aviation industry in the United States.  Until the 1970s, companies like American Airlines, Delta Airlines and TWA had monopolies on the routes they served. The airlines’ operating conditions (ticket prices and routes offered) were tightly regulated by the federal government, a situation that was not dissimilar to how the French railways operated, with the exception that the aviation companies were not public companies. At the beginning of the 1980s, deregulation in the aviation sector led to major changes in the organization of the industry. New small more flexible aviation companies staffed by younger and less well-paid people, began to compete with the big companies, which saw their revenues eroded. In response, most of them began a price war. This strategy did not produce the desired effect since, even when completely filling planes, the big companies continued to lose money. Moreover, on certain routes, business travellers used to flying frequently by plane and buying their tickets a few hours before take-off found themselves prevented from boarding because of a lack of space. Most of them, however, were prepared to pay a higher price than the one demanded by the companies. It was in this context that revenue or yield management was born, first at Delta Airlines, and then soon developed in a more elaborate form at American Airlines. 
48Until then, ticket prices were fixed within each company by a team of employees using just one means: their personal intuition. The possibilities opened up by the development of information technology in the aviation sector changed the situation however. Delta Airlines spent two million dollars on reorganizing their reservation system in an attempt to “sell the right product to the right customer at the right time for the right price” (Cross 1997: 132). Thanks to this investment, the company’s revenue rose by 300 million dollars in 1984, an increase of 100% in its annual results. More ambitious again was the system American Airlines introduced between 1985 and 1988: the Dynamic Inventory Allocation and Maintenance Optimizer (DINAMO), the first truly automated yield management system.  In three years, the extra revenue produced thanks to this system reached $1.4 billion dollars (Smith et al. 1992). Yield management was born. It is said that this new strategic weapon is transposable to all sectors faced with fluctuations in demand and where fixed costs are high.
49When Metzler decided to equip the SNCF with a yield management system in 1987, this tariff setting tool was virtually unknown in France.  No French company had yet tried such a tool, much less a company in the public sector. Some members of the GLT were thus sent on a mission to the United States. The aim of this trip was to buy a computerized reservation and capacity management system with a view to transposing it to the French railways for the forthcoming opening of the TGV Nord line. The SNCF eventually turned to American Airlines to develop the Système Offrant à la Clientèle des Réservations d’Affaires et de Tourisme en Europe (SOCRATE) (System Offering Customers Business and Tourism Reservations in Europe).  They still had to convince the company’s management of the value of purchasing such a system.
50In 1988, the SOCRATE project was presented to the SNCF senior management and board of directors by Metzler, who had to demonstrate the value to the company of changing tariff model. There was no opposition from the company’s senior management, but instead a sort of immobility in the face of an unfamiliar technique whose practical effects were unknown. Based on an initial econometric model and on a study commissioned by an international strategy consultancy, an uncommon, even unheard of, practice in the SNCF at the time, the GLT highlighted the increase in revenue that would result form a profound reform to tariff setting: “We have shown that the implementation of a yield system would enable margins to be improved by 4% in comparison to current imperfect tariff setting, which is complicated and has many exceptions. It was sufficient to offset the costs of the SOCRATE project, which nevertheless involved a profound change to the system of ticket distribution.” 
51Metzler’s social characteristics would have facilitated negotiations with company management.  The “grandes lignes” director had spent almost his entire career in the SNCF and took an active part in the birth of the TGV, the project which the company considered to be a technological achievement and an economic success. As a result he enjoyed a certain legitimacy among members of the senior management. Moreover, like the SNCF’s chief executive and three of the four deputy chief executives, Metzler was a member of the Corps des Ponts et Chaussées.  He benefitted in this regard from a form of limited solidarity that exists between members of the grands corps and graduates of certain grandes écoles (Finez and Comet 2011), but also from symbolic power allowing him to propose redefining the train ticket tariff setting system. The development of the SOCRATE project (1988–93) coincided, moreover, with the arrival of a new generation of executives at the top of the SNCF—with more diverse educational profiles, often graduates of ENA or business grandes écoles, sometimes coming from the private sector—who were particularly in favour of the company’s financial success (Finez 2012).
52The yield management system introduced on the LGV Nord high-speed line in 1993 was in part similar to the TGV Paris–Lyon tariff model. Train ticket prices were calculated based on a minimum price for each rail journey depending on the captiveness of the micromarket. A supplement was added to this base tariff when travelling in peak periods along with an obligatory fixed reservation fee. Yield management also introduced quotas of reduced-price tickets, the scale of which varied depending on train occupancy levels, the time of ticket purchase and prices offered by competitors. Price levels also depended on a series of criteria: whether the ticket was refundable or exchangeable (with or without cost), possession of a railcard, etc. Finally the SNCF practised a policy of overbooking: meaning selling more seats than available on a train, with the knowledge that some passengers would not turn up when the train leaves.
53Price setting relied on the work of a team consisting of around fifty people in 2012. Located in the business district of La Défense, the yield managers in the Centre d’Optimisation Commerciale (COC) (Commercial Optimization Centre) adjust ticket prices in real time, using increasingly complex computer algorithms. Their work entails optimizing train occupancy rates while ensuring that ticket prices are as close as possible to the maximum sum each consumer is prepared to pay to travel. This means giving priority to the most affluent customers, in particular those travelling for business, without, however, excluding revenue from other passengers (young people, families, people on lower incomes, etc.). Yield management, the tariff setting technique that soon spread across the whole TGV network and then to Corail (non-TGV national) trains in the 2000s, thus aims to maximize the company’s revenue.
54It is hard to gauge the extra margins that yield management actually produces. What the director of the COC says about the strategic role of real-time tariff setting for the company is however illuminating: “Once introduced, yield management is not expensive to run … When you make one euro more, it’s practically an extra one euro margin. This euro turns almost entirely into profit! You can see the lever effect, can’t you? One percentage increase in turnover represents about 50 million euros for the TGV. Although one point might seem small in terms of the company’s margins, it’s actually enormous. … I’m not going to tell you how much more yield management extracts. I can however tell you that it’s commonly accepted that, thanks to these techniques, operators can increase their revenues by a range of 3% to 10%. Then it depends on the techniques used and the maturity of the market, brand image, etc. … As for us, we’re probably number one in the world in the railway sector, ahead of Virgin and Amtrak.” 
Yield management as a new principle of fairness
55When yield management was introduced, the trade unions protested against what they saw as “class tariff setting.”  The new tariff system effectively prevented less wealthy customers from travelling at peak times, thereby encouraging social divisions. Some unions, like the CGT, also emphasized that ending the kilometric scale amounted to undermining the “historical foundations” of the principle of fairness in rail transport.  The president of the SNCF at the time, Jacques Fournier, asserted in contrast that the tariff innovations were part of the huge changes that were affecting all public service utilities (e.g., electricity and telephone) and that the principle on which yield management is based was “analogous to that which had presided over the introduction of the three-colour calendar” (Fournier 1993: 51–2). There was thus no reason to be worried about the effect of introducing SOCRATE.
56It is true that the incremental nature of the innovations and the hybridization of price models that resulted might give the impression of a sort of continuity between blue–white–red tariffs and yield management.  In reality, the foundations of the two tariff setting models differ profoundly, as much in terms of the nature of metrological operations enabling the price of train tickets to be fixed as from the point of view of the management objectives they serve. The adoption of yield management signified the abandonment of a price calculated on the basis of costs in favour of tariff setting based solely on the utility passengers accord to the transport service. Moreover the new tariff system was meant to make the company a profit and was not simply aimed at balancing the company’s books. A recent brochure published by the Centre d’Optimisation Commerciale describing a revenue management analyst’s work clearly states: “At the SNCF, your role is to optimize receipts and the occupancy of a portfolio of trains in your charge (15 to 20 trains per day, or a turnover of 150 million euros on average).” Yield management was thus a change in tariff setting paradigm and not simply the automation of price discrimination as it existed in the 1970s and which would have been made possible by developments in information technology.
57It would, however, be wrong to believe that the introduction of yield management in the SNCF and its expansion over the course of the last two decades could have been done without taking into account the requirements of social justice and debates around its compatibility with the principle of public service. As with marginal-cost tariff setting, the adoption of this new pricing system was accompanied by efforts at moral justification. Yield management “is not, in essence, contrary to consumer interests and public service, quite the contrary,” stated a recent parliamentary report (Assemblée Nationale 2008: 39), confirming the opinion of the Conseil d’État (1993) which recognized the need for the SNCF to abandon the kilometric relationship and to vary prices in order to “improve the profitability of the service depending on the particular characteristics of its clientele.” Using a price–quantity demand curve, it is easy to show that tariff discrimination potentially allows the satisfaction of more consumers than a single price (tariff equalization) or moderate price discrimination (three-colour tariffs). In doing so, yield management can contribute to a form of increased wellbeing for individuals in society.
58Such analyses are similar to the position supported by the SNCF general management and the GLT. The president, Jacques Fournier, claimed that as long as ticket prices “stretched over a sufficiently wide range, … there is no basis for] criticizing them from the point of view of the fight against social inequality” (Fournier 1993: 52). For Metzler (1988), the SNCF was a productive organization, which like other public and private companies, must adapt to changes in behaviour and mentalities: “customers are changing. They’re more demanding, they’re better consumers” and is tending even to behave like a “perfect homo œconomicus.” It is therefore necessary for the company to adapt and to seek to satisfy passenger demand in a rational way—with the understanding that we are talking about “economic rationality” here. There is an element of an old principle from the liberal political economy in this discourse (Smith 1776), which would have it that the meeting of particular interests (the passengers’, who want to minimize their expenditure, but also the SNCF’s, which must seek to make a profit) contributes to the general interest. Because it really is the general interest, defined as the satisfaction of most, that is at issue: Metzler, as a Polytechnicien and member of the Corps des Ponts et Chaussées, was one of its strongest champions. 
59In this respect it is interesting to note that yield management is an application of utility tariff setting theory formulated by Jules Dupuit in the middle of the nineteenth century (Dupuit 1844). According to this liberal Ponts et Chaussées engineer, differential tariff setting, which consists of requiring each network user to make the maximum contribution they are willing to pay, enables the needs of the greatest number of people to be satisfied.  At the time it was set out, Dupuit’s utilitarian theory had little influence on the state engineers in charge of regulating the railways (Ribeill 1993). The recommendations stemming from it were effectively judged to be against their economic morality. They considered that tariff setting should be based on running costs in order to conform to the principles of measurement in industrial engineering (Grall 2003: 386). The adoption of real-time tariff setting by the SNCF in 1993 was thus an implementation of an old principle of utilitarian fairness in terms of access to transport, a “higher morality” (Vatin 2002) that had remained just a theory for nearly a century and half.
60The paradigm shift is also, however, part of the historical process of increasing autonomy of the SNCF from its masters. It is true that the state is still supposed to control the company’s tariff setting policy: prices must not exceed certain fixed bands (Perennes 2012). Too great a gap—which remains at the discretion of the public authorities—between the most expensive and cheapest tickets would be judged contrary to the principles of public service. In 2008, the price of the same physical seat on a TGV could vary between 1 and 3 (Assemblée Nationale 2008: 39). Recent changes in the SNCF’s marketing policy, encouraged by the adoption of a decree (Assemblée Nationale 2011) granting the company greater tariff setting freedom, were accompanied however by an increase in the spread of prices.
61Doubt could also be cast on the whether the state owner has a real tariff regulation policy, in particular concerning the fluctuation of prices between the minimum and maximum bands. In the opinion of a senior civil servant who was in charge of monitoring prices in the Ministère des Transports, the state should not interfere with tariff setting policy on the SNCF “grandes lignes.” And what is more, it does not do so: at the end of the 2000s, “we used to receive an Excel document. We would open and glance at it, that’s all. We wouldn’t check anything. … I don’t see why the state should control prices either. For me, a large part of the SNCF’s operations, like the TGV, are like the private sector. … The state should behave like a shareholder.”  What this senior civil servant says is as much symptomatic of the changing attitudes among the elite in this sector as it is of the consolidation of the contractual relations the rail company has with the state. The public authorities, having become “state shareholders” (Delion 2007), encourage public companies to generate profits. Thus they can limit the financial support they offer, even reap “dividends” from those that make a profit. Between 2008 and 2013, the SNCF has thus paid nearly 800 million euros to the state.
63Three structuring paradigms emerge through the various reforms to railway tariff setting introduced incrementally since 1938. The first model, which pre-existed the SNCF, fixed ticket prices according to distance travelled. Then in the decades following the war, the uniform per-kilometre tariff was slowly abandoned in favour of a new principle: the price of a ticket should henceforth approach operating costs as much as possible. Finally, in the 1980s–90s, marginal-cost tariff setting was replaced by yield management, a price system based on the utility passengers attribute to a transport service.
64The SNCF tariff changes resulted from the work of a few entrepreneurs in economicity who acted in political, competitive, social and also technical contexts that defined the space of possibles for the tariff formulae they proposed. Without developments in road and air transport in the second half of the twentieth century and the opening to competition from European railway markets in the 1980s, without a transformation of the mindsets of state elites and the process of management rationalization in public companies, without changes to power relations within the senior management of the SNCF, the question of the financing of French railways would have been posed in a completely different way and the responses given would have been of a different nature. The motivations and capacities for action of the price setters should be sought in their social characteristics. Tariff innovators were “allogeneic” actors and “in-house” managers whose profiles and ideas were relatively specific with respect to the norms and images of the SNCF of its times. Boiteux and Hutter were engineers educated in mathematical economics and convinced by the Allaisian theory of social efficiency, Metzler and the GLT were imbued with the values that prevail in the private sector. The peculiarity of the social profiles of these entrepreneurs in economicity—which, a priori, limited their influence—was compensated for by the ties they had with certain executives or by the fact that they themselves occupied elevated positions in the sector.
65This historical sociology of tariff setting emphasizes the crucial role of economics in the adoption and justification of new price formulas. The SNCF price setters used economics senso lato in order to demonstrate that their forms of economicity benefited passengers, the SNCF and the state. Originally, the per-kilometre tariff was based on a classical principle of market fairness: it is fair that all network users pay an identical price for the same service. In his theory of social efficiency, Allais highlights the unintended consequences of a price that is proportional to distance and suggests adopting marginal-cost tariff setting. This was intended to guide consumers to consumption that was better suited to the general interest and promote a balanced SNCF budget. Backed by developments in computing, yield management made possible tariff setting based on the differing utility that passengers accord to the same service. Yield management enabled the SNCF to broaden its customer base and reap additional profits each year. These two successive redefinitions of the general interest were accompanied by a redefinition of the principle of public transport services.
66Twenty years after having introduced SOCRATE, reforms to tariff setting in the French railways are far from over. The changes to tariff setting for national passenger transport have been retraced in detail. We now need to know how price systems have evolved and will evolve at the same time in other railway markets (rail freight and international and regional passenger transport), which also faced in recent years with increasing intra- and intermodal competition. Furthermore, the creation of the Réseau Ferré de France (the French Railway Network) in the context of European liberalization of the railway sector now raises the question of tariff setting for infrastructure use. Historically, access to the network was free and was based on a simple principle: passenger trains had priority over the transport of goods. The gradual opening of markets however has dramatically shifted the management of train paths from the world of engineers to that of economists (Crozet 2011: 222–4). Two economic “truths” confront each other today. Some experts believe that rail tolls should cover all costs and that the prices for train paths should be fixed in a way that selects traffic depending on the disposition of users to pay. Others, more interested in competition being effective—in accordance with recommendations from the European Commission—, assert that setting tariffs to cover all costs would have a dissuasive effect on new entrants that might compete with the SNCF. This is the reason why it would be preferable to set tariffs at marginal cost, the inherent losses being covered by the state.
67The history of railway tariff setting is certainly a history of increasingly complex price models. Yield management profoundly changes the conditions for the meeting of supply and demand since potential customers are confronted with the Centre d’Optimisation Commerciale’s powerful algorithms in a situation of information asymmetry. Passengers seem relatively comfortable with this situation and have internalized the fact that it is possible to travel on the network at a cheaper price as long as they spend time in front of the computer, are sure they will not change their minds (the cheapest tickets are non-exchangeable and non-refundable), are flexible about travel times, anticipate the probable changes in ticket prices from one day to the next, etc. Therefore, has the yield management system—whose operation is based on the standard hypothesis of homo œconomicus—had “theory effects” (Garcia 1986) on passenger behaviour? To what extent has the necessary computational logic needed to buy a ticket transformed passengers into rational consumers? In the absence of data allowing us to make sound hypotheses, such questions must remain unanswered for the time being. However, they offer a stimulating research perspective for those who want to understand how demand reacts and is restructured in the face of changes to the services offered by the SNCF, of which price is one of the components.
[Translator’s note] I have chosen to translate the French word “tarif” and derivations of that word used in the context of this article as “tariff” rather than translating it as “fare,” “rate” or “ticket price” as more usually used in English when referring to pricing in the rail sector. The justification for doing so is explained by the author in the “The study of SNCF prices as an analysis of a state-controlled tariff” section below on pages 6–7.
As part of its business strategy, a railway company may also close the least profitable lines or stimulate demand through an appropriate marketing policy.
For the sake of clarity, this study is based on the national transportation of passengers, what is now called the “grandes lignes” (“mainlines”). I have left aside a study of the issue of setting tariffs for local, regional and international passengers. In the section devoted to marginal-cost tariff setting, however, I look into the calculation of prices for the rail transportation of goods.
Kuisel (1981) offers an overall picture of the French economy up to the 1950s. Bezes (2009) analyses reforms in the French administration between 1962 and 2008 (rationalization of the management of public finances, the introduction of new public management, etc.). Denord’s (2007) work on neo-liberalism and its advocates allows us to analyse the debates on tariff setting in the SNCF in light of the ideological conflicts in France between 1930 and 1980.
A sector can be labelled a natural monopoly when there are significant economies of scale and where, as a consequence, it is less costly for the production of all the goods or services to be entrusted to a single operator.
A cost effectiveness entrepreneur is an actor who attempts to transform the practices and economic performance of individuals he manages in such a way that they conform to his economic morality, in particular in terms of fairness and economic efficiency. A thesis currently being prepared on the electricity sector studies forms of cost effectiveness at work in tariff reform at the EDF (Yon 2012).
The grandes écoles are public and private higher education institutions that recruit their students through highly selective competitive examinations and that educate France’s future political, administrative and economic elites. The best known grandes écoles are ENA (École Nationale d’Administration [National School of Administration]), the École Polytechnique (science and engineering), Sciences-Po (political sciences) and HEC (École des Hautes Études Commerciales [business school]).
The Revue Française de Sociologie (Chiffoleau and Laporte 2004; Éloire 2010; Chauvin 2011) and the journal Sociologie du Travail (“La qualité” issue, 2002; Barrey 2006; Anzalone 2009) have clearly led the debate on this subject. Note, however, that the sociology of prices cannot be reduced to price setting alone. There has been some research, for example, into operations that precede monetary valuation. Evaluating an object, giving it a price, means making it a commodity, which is not always obvious, especially when it comes to human life (Zelizer 1985) or “nature” (Fourcade 2011). Other studies seek to explain price volatility (Baker 1984), others to understand the structure linking prices to each other in a given market (Chauvin 2011).
The collected work, The Worth of Goods. Valuation and Pricing in the Economy (Beckert and Aspers 2011) is symptomatic of the special nature of goods looked at by the sociology of prices. Most of the case studies presented look at “aesthetic” markets (wine, fashion, contemporary art, etc.) and financial markets.
Recent studies on the catering market, based both on H.C. White’s sociological theory of markets and post-Keynesian economic theory are notable exceptions within the French literature in this respect (Dallery et al. 2009; Éloire 2010).
To be more precise, the diachronic analysis of price setting proposed is a three-stage process: first the development of tariff models, then their questioning and finally their replacement with new models.
The word “tariff” stems from the Arabic word ta’rif meaning “making known.” A tariff is therefore a defined price, made public and for which there is no uncertainty about the sum to be exchanged.
In this article I do not use the principle of “performativity” (Callon 1998; MacKenzie and Millo 2003; Mackenzie et al. 2008). Indeed, apart from defining the notion in a broad sense by pointing out the principle according to which knowledge and representations of the economy are transformed into practices, this study looks no further. I believe we can speak of economic performativity when the application of a theoretical model (for example the model of perfect competition), some of whose hypotheses (such as the hypothesis of the economic rationality of agents) are unverified empirically, transform economic reality to the point that these hypotheses become real (agents act as homines œconomici).
This seems particularly true for the introduction of marginal-cost tariff setting. In contrast, for the most recent and thus most secret and least known period, it is harder to determine the role of the “social capital” of the advocates (Bourdieu 1980b) for the adoption of yield management.
Among the economists and senior civil servants some belonged to both groups, for example Maurice Allais, Louis Armand, Gabriel Ardant, François Bloch-Lainé, Jean Fourastié, Jean Monnet and Alfred Sauvy.
For a historical sociology of the effects of the mechanization of information handling on contemporary economies and societies, see Gardey (2008).
While, in the name of equality of treatment, per-kilometre tariff setting existed at the time of private companies (around 1840–1938), it would be wrong to believe that this was easily achieved. The state imposed a legal tariff in the mid-nineteenth century, but companies invented all kinds of strategies aimed at circumventing it (Ribeill 1993). From the 1890s, the state relaxed the tariff framework and allowed companies to offer reduced tariffs in certain circumstances: return tickets, season tickets, tickets to tourist destinations, etc. (Caron 2005: 295–307).
Cahier des Charges de la Société Nationale des Chemins de Fer Français (art. 2: 356) approved by decree (Présidence de la République 1938).
SNCF (1971: 5–6).
The “indicateur Chaix” is a directory that contains a list of train routes and timetables, as well as the distances between different stations around the country, enabling the price of a train ticket to be calculated in advance.
This formula, like those that follow, reflects the “formulae” scribbled down by SNCF price setters during the interviews they gave us.
Note that geographic uniformity has never been fully implemented. The length of some lines crossing the Massif Central was artificially increased in order to take into account the higher cost of transport in this region. This hidden practice would have been carried out with the complicity of the state owner (Dumont 1998: 25).
On leaving the École Polytechnique graduates are ranked. The best students are incorporated into the technical grands corps of the state, in particular the Corps des Mines and the Corps des Ponts et Chaussées, which are the most prestigious. These state engineers, who are destined to hold high positions in the administration and large public companies, maintain an “esprit de corps”: they share a set of representations and develop forms of solidarity.
Account deficits in the railways first occurred in the 1910s, gradually disappeared in the 1920s, then returned and worsened from the end of the 1930s onwards. In his study of the SNCF’s trading accounts for 1952, Roger Guibert, a senior official at the SNCF, estimated the deficit to be 32.3 billion francs. For him, this confirmed the SNCF’s inability to face up to competition from other modes of transport as a result of a lack of coordination between road and rail and an inappropriate tariff policy (Guibert 1956: 173–84).
This expression describes engineers belonging to the technical corps of the state who are interested in economic theory. They consider that economic calculations are, in the same way as knowledge and technico-physical issues, an essential aspect of the good management of the country’s productive activities.
The recommendations from Allais and his “students” were not however limited to rail tariff setting. They also related to railway workers’ pay and the internal organization of the SNCF (Finez 2013b).
A fellow pupil at the École Polytechnique and the École des Mines, but also a friend of Allais, Hutter is regarded as “the SNCF’s transport policy economic thinker” (see his biography in the Annales des mines) having spent his entire career there. As for Boiteux, he was Allais’s assistant at the École des Mines. He spent a short time at the SNCF in 1947 and was then recruited by EDF in 1949.
Since the nineteenth century, an ad valorem tariff for rail goods transport prevailed: the price for transport was based on a per-kilometre tariff qualified pro rata by the value of the goods transported.
According to Delefortrie-Soubeyroux (1961), at the end of the 1950s, of the 159 “directors” of the SNCF responding to a large survey of directors of French industry, 94% were engineers and 98% spent their entire career within the railway sector. They were mostly graduates of the École Polytechnique (around 70%).
The “économie de forces” is a notion created by Grall (2003) to describe the mode of thinking of engineers in the Corps des Ponts et Chaussées who share a particular perspective on economic efficiency. According to these engineers, a productive activity is organized efficiently when the energy expended to carry out a given task is the minimum, in other words when mechanical efficiency it at its maximum.
The École Nationale des Ponts et Chaussées is a grande école that educates engineers destined to fill the highest positions in the public works and transport sectors. In particular it has responsibility for completing the education of Polytechniciens of the Corps des Ponts et Chaussées.
Jacques Rueff was an engineer-economist and influential senior civil servant during the Fifth Republic. In particular, he played a major role in the development of neo-liberalism in France and in implementing economic reforms in the country. Louis Armand was an engineer and senior civil servant who spent his career in the railways. Specifically he was director general (1949–55), then president (1955–58) of the SNCF. He is considered to be one of the major figures in the development of the sector.
Here an allusion is made to the work on tariff setting carried out in 1947 by Boiteux and Hutter.
Ministère des Travaux Publics, des Transports et du Tourisme (1960: 6).
See Assemblée Nationale (1961: 3783).
Simon Nora is a senior civil servant who played a central role in the economic “modernization” of France. He is a graduate of ENA and a member of the Inspection Générale des Finances (Inspectorate General of Finances), which is the most prestigious administrative grand corps.
Describing the influence he believed he had on instituting the first reforms to passenger tariffs, Boiteux noted: “Fifteen years to receive a professorship, fifteen years for my students to come to power [that is in the Ministère des Transports and the SNCF], that’s a good thirty years from idea to realization. That’s a lot.” (Boiteux 1993: 47).
This is based on a study of the education and career paths of all the members of the SNCF’s senior management between 1938 and 2012, about 80 directors. A more detailed study that was the subject of a presentation (Finez 2012) is being published.
Revue Générale des Chemins de Fer (1970: 105). The article does not detail the reason behind the SNCF fixing the surcharge at this level.
Interview of 22 November 2011 with a former director of the SNCF.
Since at least the 1950s, SNCF engineers and managers had taken trips abroad (to Germany, the US, Canada and Japan in particular) and imported and imitated certain national railway company strategies (Finez 2013a).
Letter of 16 April 2012 from a former CFDT union official at the SNCF.
By setting the price on the route rather than the length of the journey, the company also implicitly highlighted the limits to the per-kilometre tariff, which charges passengers for the network’s inconsistencies.
CFDT-FGTE Cheminots, 1981, TGV. Pour les usagers, le prix à payer sera plus lourd, July–August leaflet.
Interview of 22 June 2012 with a former CGT Cheminots (General Confederation of Labour Railway Workers) trade union official.
Although, from the 1980s onwards, the SNCF had abandoned the kilometric scale for TGV lines, the base tariff for each journey has remained state regulated today. The maximum price for a 2nd class ticket, in particular, cannot exceed a certain level, still calculated today based on the distance between the two stations. In contrast, first class tariffs are completely unregulated (Perennes 2012).
A graduate of the École Polytechnique and member of the Corps des Ponts et Chaussées, Metzler joined the SNCF in 1967 and built a career working on the TGV Sud-Est project he managed. Between 1983 and 1986, he left the SNCF to spend a short time at Schneider for whom he managed various projects abroad, before returning to the public company. His experience of private industry would have a decisive influence on his strategy for the commercial management of the SNCF (Dumont 1998).
Interview of 21 November 2011 with a former GLT manager.
The history of the birth of yield management is well documented (Smith et al. 1992; Cross 1997; Talluri and Van Ryzin 2005; Boyd 2007). In contrast, despite having a very supportive sociological study entirely devoted to it, there are still few studies on the subject. In his work, La sociologie économique, Steiner devotes a section to this particular price scheme (2011: 91–2). In addition, a sociological thesis on the subject is ongoing and a working paper has recently been published (Pigounidès 2013). Note also Mitev’s (2004) work dealing with difficulties implementing the SOCRATE reservation system at the SNCF—yield management being one of its components—from the perspective of the sociology of management of technology and information.
Note however that yield management did not appear ex nihilo. It was part of a continuity of tariff strategies that emerged in the United Kingdom in the 1970s: in response to a change in competition regulations in the aviation sector, mathematical demand forecasting models were developed at the British Overseas Airways Corporation (BOAC) (Pigounidès 2013).
Most of the prices for ticket sales were set by algorithms, which enabled tariff setters to concentrate on flights where demand was difficult to model.
The first thesis on yield management was written at Massachusetts Institute of Technology (MIT) by Peter P. Belobaba (1987). The first book on the subject appeared two years later, written by two GLT managers (Daudel and Vialle 1989). It was translated into English in 1994.
SOCRATE integrated THALÈS and ARISTOTE, two key software packages for yield management tariff setting. THALÈS optimized the allocation of places relying on ARISTOTE, the archive and study tool for information on past traffic.
Interview of 21 November 2011 with a form GLT manager.
Interview of 23 November 2011 with a form GLT manager.
The fourth deputy chief executive was a graduate of the École Polytechnique and ENA.
Interview of 27 September 2011 with the director of yield management in the SNCF.
Interview of 22 June 2012 with a former trade union official from the CGT Cheminots.
Here I refer in particular to the SNCF’s pricing policies in the 1980s, which were still imbued with the marginal cost paradigm, but had begun to take an interest in customers’ differentiated means to pay.
Interview of 22 November 2011 with Metzler.
The engineer thus stated: “I spoke above about the publisher who managed to sell his book at different prices, and the impresario who managed to fill his theatre with a combination of prices. I would tell a railway operator: do as they do” (Dupuit 1849: 237).
Interview of 19 June 2012.