1 WHILE ECONOMIC PRACTICES almost always include calculation, forecasts, and, therefore, time, one practice is seen by some authors as the cornerstone of the social bond and has a profoundly temporal dimension: credit. [1] In recent times credit has been the subject of much interest throughout the world's universities. [2] This article approaches the issue through the question of prediction, focusing on the moment credit is granted to understand the connection and the reciprocal influences between lenders’ conception of the future and credit granting methods, in particular credit scoring.

2 Credit scoring consists in assigning a score to credit applicants that describes the level of risk they entail for the lender. To work out this score, lenders need to possess information on the applicant and data on previous borrowers based on which they can determine the statistical link between individuals’ characteristics and risk levels. Three temporalities combine to form a score: the future, since it is the likelihood that the loan will be repaid that is rated; the past, since the score is established based on previous borrowers, but also on the behavior and past experience of the applicant; and the present, the moment credit is granted, when the lender accepts or refuses to lend and decides on the terms and conditions of the loan.

3 The score of a given individual is unique, but predicts nothing about non-repayment by that individual. The lender expects that a proportion of its borrowers will fail to repay and, based on this proportion, sets an interest rate that will cover losses. [3] Non-repayment is experienced as a major event by an insolvent borrower (with consequences such as loss of house, various fees, legal troubles, and so on) yet not for the bank, since it will have already anticipated a portion of losses. One of the roles of finance is to make goods and services liquid, in other words to remove them from their fixity, their local confine, the links they may have with individuals, and, potentially, regulations that can curb the speed of their circulation. [4] Credit scores have a great ability to liquefy the market as they provide a synthetic piece of information that enables all actors to have the same standardized knowledge of a loan's value. But the credit score still needs to circulate, and here the comparison between France and the United States is interesting: in France it is an in-house tool for banks, whereas in the United States it is available for public purchase.

4 While anthropologists and historians have described idiosyncratic forms of credit rooted in social structures, [5] the development of mass credit and its extension into derivative products sold in financial markets have led to major change in the ways in which risk is calculated. Looking beyond measurement techniques, ho wever, it is the very use of methods to predict and forecast the future that has been transformed.

5 We will draw on the distinction made by Luc Boltanski in On Critique between “reality” and the “world.” [6] Reality is that which is set in order and which “appears to hang together,” whereas the world is “everything that is the case.” [7] Institutions see to it that reality is created and maintained, while keeping constant watch for any sudden intrusion by the “world.” These two notions echo the distinction made by Frank Knight between risk and uncertainty. [8] Observing statistics—in particular those used by insurers and lenders who, acting under the threat of default themselves, must precisely calculate the financial risks they incur—Knight showed that there are two types of future events. Some events are probabilizable and can be used as a basis for financial calculations (for example, the mortality rate by age). These constitute risk. By contrast, others can only be anticipated by specialists (such as earthquakes or the sudden outbreak of an epidemic). These events constitute uncertainty. The trick for companies whose profits are based on anticipating the future is knowing how to transform uncertainty into risk, or, in other words, finding ways to calculate mathematical probabilities for events that hitherto seemed random. Since the nineteenth century, an increasing number of uncertain factors have entered the domain of risk and have become potentially insurable or objects of speculative bets. Seen in this light, Boltanski’s “reality” is Knight’s risk, the “world” is uncertainty.

6 In Énigmes et complots, an analysis of crime fiction, and therefore of the threat of uncertainty, Boltanski brings to light the efforts of modern states to stabilize “reality” and to expel the “world.”  [9] The device studied in this article also features at the heart of such questions. We will reflect on the type of reality that credit scoring seeks to create and, in particular, the decoupling between how the event, the “world,” and its materialization are experienced by borrowers (or even bank employees who are in contact with them), as well as their neutralization in financial instruments and through liquidity, which detaches them from individual experience.

7 By comparing the way in which statistical methods of risk assessment are used in France and the United States, and the type of credit market that results, we will show that the discounting of risk in a score has the consequence of transforming what could initially be viewed as an event—the non-repayment of a loan in the history of given borrower—into an element of reality, as if time could be erased and that which “is the case” could be ignored.

1. Credit Scoring in France: Classical Analysis of Statistical Risk Assessment Methods

8 In contemporary France, any application for consumer credit from a bank or credit institution is assessed using a score. This score is usually ternary—green, orange, or red—and indicates whether the application of the customer in question should be accepted, rejected, or referred to a superior. In specialized credit institutions [10], numerical scores grade customer behavior: a new customer has a low score, which increases the more the customer borrows and honors repayment commitments, which then opens the possibility for further borrowing. While the customer is unaware of this score, he or she may then receive offers for additional credit or positive responses to new applications. On the other hand, the score can decrease in two cases: repayment problems and recurring signs of difficulty.Credit scores are an in-house tool for banks and, in either case, are unknown to customers. They separate customers between those whose credit applications should be accepted and those to whom credit should be declined. The threshold is determined by the banks: the credit score does not say whether an individual will repay or not, but rather the level of risk that he or she represents. It places the individual in a group among which, for instance, five percent or two percent will not repay. [11] Depending on the profit they expect to make, lenders then choose the level of risk they are prepared to incur (generally less than five percent non-payment, and usually around two or three percent). In France, their interest rate—and therefore the price of credit—is capped by a usury rate, which prevents them from lending to excessively risky groups, since profit on loans actually repaid would be insufficient to cover losses.

9 Banks use the data they possess on the applicant to calculate his or her score. This includes administrative data, civil status, occupational data, wealth, and so on. In the event whereby the applicant is already a customer, his or her repayment history can also be considered. The French data protection agency (CNIL) prohibits these scores from being shared among banks and credit institutions. In any case, competition between lending establishments would not favor such crucial information being shared. The credit score is private information. This is also a significant impediment to foreign banks that wish to establish themselves in France: they do not have access to any scores or credit history available for public purchase. Their only available solution is to acquire an entire bank, as was the case with HSBC. [12]

10 Lenders’ use of a standardized scale to decide whether to grant credit is not a recent phenomenon. In a 1963 publication, Luc Boltanski, Pierre Bourdieu, and Jean-Claude Chamboredon illustrated a highly standardized selection process in which the ability to repay was assessed by an objectification of the occupational and social situation of applicants. [13] But such assessments did not amount to scoring, because acceptance criteria were not based on statistical evaluations of past loans, but rather on criteria established by the lender.

11 French banks gradually incorporated credit scoring into their procedures in the 1970s. The first mention of the use of scoring in the pages of Banque, the leading journal for professionals and institutional players in the banking sector, concerned a study carried out in 1971 that applied solvency scores to 2,200 customers. [14] The matter was not explored in depth, however, until the latter part of the decade: computerization and the proliferation of forms of credit that had become available to individuals (credit cards, personal loans, consumer loans) made rating systems to standardize and speed up response times both necessary and possible. The credit score opened the path from “specialization” to “consolidation,” [15] and made it possible to respond rapidly to loan applications. Nevertheless, the journal’s authors underlined that the human assessment remained fundamental and that these ratings were merely a support.

12 Technically, credit scoring requires the possession of data concerning old loans, processed to predict the probability of repayment of future loans. Banks therefore need to acquire “data warehouses,” which are enormous machines in which all information concerning their customers is stockpiled (and which is also used for other services, such as marketing in particular), as well as software developed by specialized companies to create the scores and translate them into interfaces on the computers of the employees who liaise with customers. The largest banks were the first to make this costly investment, and it was not until the Basel Committee issued its recommendations in 2004—the so-called Basel II accord—that the last remaining banks adopted the technical tool. [16] Basel II transformed the methodology for calculating the amount of capital a bank was required to put aside as a proportion of its loans. [17] In Europe, this was fully transposed into law by the European Parliament Directive 2006/49/EC. Very broadly speaking, this share of capital was previously a percentage of loans granted by a bank depending on the type of loans. Henceforth, it was the level of risk assigned to borrowers that determined the level of capital that a bank was required to hold. If the banks did not assign scores to all their customers, they would have to use a system based on an evaluation by credit rating agencies, which would result in higher capital requirements.

13 The development of credit scoring in France quite closely mirrors descriptions of the same phenomenon in the United Kingdom made by Andrew Leyshon and Nigel Thrift: a spatial shift in the decision-making process made possible by the accumulation of data on customers and desirable in the eyes of management with a view to standardize practices. [18] Another way of reflecting on this technical tool is to observe, along the lines of George Akerlof’s market for lemons, the institutional conditions that allow a market based on scoring to develop. [19] Information technology is, of course, needed to centralize data and to establish calculations, but “reality” also needs to have been stabilized—it must be possible to measure the future and this measure must be able to be made on the basis of past data. In this respect, Alya Guseva’s article on Russia, in which the author examines the development of the Russian credit card market, is very interesting. [20] While the first credit card was released in Russia only in 1989, by 2009 the country counted 125 million cards. This growth is all the more spectacular considering that Russia did not have a tradition of consumer credit, or for that matter one of consumption. Russian banks possessed no credit history on which they could establish scores. First, they distributed these cards through direct contact with customers. Second, they offered them to the executives of major firms. Third came the implementation of “salary projects”: all employees, from the sweeper to the CEO, were given access to an account and a card. The final stage was the release of “express loans” in large stores. These were standardized and were not accompanied by any social system of coercion. The author shows that banks used existing social structures and control to develop their market. This institutional interpretation was also adopted by Leyshon and Thrift, who emphasized the bond between credit scoring, labor organization, and the distribution of power within banks.

14 This institutional approach could, however, lead one to believe that credit risk already existed, and was merely awaiting calculation. Michel Callon and Fabian Muniesa have shown on the contrary that any calculation requires the prior creation of a calculable space, and objects need to have been made calculable: “Calculation starts by establishing distinctions between things or states of the world, and by imagining and estimating courses of action associated with things or with those states as well as their consequences.” [21]

15 The history of credit cards in Russia also suggests that in order to create “reality,” the financial institutions first had to test the “world”: uncertainty first needed to be addressed in order to develop a credit history and establish scores. This is also what the first American banks did to distribute credit cards. They organized “drops” of pre-activated cards to thousands of random letter boxes around the country. [22]

16 The goal of credit scoring is to integrate all temporalities, in other words to anticipate all eventualities of that which “is the case,” since statistics do not predict non-repayment by a given borrower, but rather the proportion of bad loans, including if the institutional conditions to prevent fraud and opportunistic behavior are in place. The evolution of credit scoring in the United States is testament to its capacity to persuade both borrowers and lenders that the world and events can be controlled.

2. Credit Scoring with FICO

17 There are profound differences in the way the credit market operates between the United States and France. Observers often compare outstanding credit in the two countries and conclude that the French are overcautious and that the Americans are almost naturally endowed with a “credit culture.” In 2010, for example, average outstanding consumer credit per capita was €2,410 in France, compared with €5,886 in the United States. But while the United States may appear to be the home of mass credit, this is less due to the amount of outstanding credit itself than to the way in which the market is organized and lenders interact with potential borrowers. A comparison between the two countries therefore needs to consider criteria for credit lending and, more specifically, the use of credit scoring.

Conclusion

28 Subprime loans constituted sources of deconstruction of the future for borrowers, for the world of finance, and for the public debt of governments. The problem was rooted less in erroneous expectations of risk than in the transformation of the way in which lenders’ related to risk: they no longer sought to avoid it, but to incorporate it into their activities and place it on the market.

29 Risk was thus stripped of any temporal or experiential significance: the occurrence of defaults experienced as events by individuals formed part of banks’ expectations, but as elements of calculation. Yet those who file for such bankruptcy are treated in the public space as victims of an “accident” [42] and candidates for social security. Commentary concerning such people has ranged from compassion to lectures in morality. On either side of the Atlantic, it is a matter of social expectation that each borrower repays his or her debts. Lenders themselves have debt-collection services and put in place procedures based on personalization. Joe Deville has described the efforts of collectors to create attachments and affection with insolvent borrowers. [43]

30 The selection of borrowers by credit scoring is, therefore, indeed a form of exploitation in the Marxist sense of the term: the risk—in its usual sense, namely the occurrence of an event—is entirely borne by the borrower, who pays in the form of high interest rates and, especially, the real consequences that he or she may face. Meanwhile, for the banks this risk appears (appeared) to be fully secured.