Social sorting as ‘social transformation’: Credit scoring and the reproduction of populations as risks in South Africa

Introduction

[Credit] scoring is starting to get a bit over-rated ¹

On 4 December 2003, in the White House’s Roosevelt Room, US President George W. Bush signed into law the Fair and Accurate Credit Transactions Act. Minutes before doing so, he proclaimed, ‘The bill I’m about to sign will help make sure that hardworking, law-abiding citizens are treated fairly when they apply for credit’ (cited in PR Newswire, 2003). One measure in the pursuit of this ‘fairness’ is Section 215 of the Act that requires state agencies to study the impact of credit scoring on ‘geography, income, ethnicity, race, color, religion, national origin, age, sex, marital status, and creed’. In fulfilling this requirement, a report from the Board of Governors of the Federal Reserve System (2007: S-4) found that ‘blacks perform worse than other racial and ethnic groups with similar credit scores’ (see also Nelson, 2010). However, what we see from this approach of promoting ‘fairness’ is a tension between, on the one hand, legislating the eradication of ethnicized and racialized terms of access to credit and, on the other, sustaining the relationship between ethnicity, race and credit scoring. The latter relationship not only fails to move away from conceptualizations of financial delinquency ² as an inherent ethnic and racial trait, but places financial responsibility configured in this way on the shoulders of profiled consumers rather than of those profilers who arguably benefit.

The scale of the financial crisis that followed shortly after the passing of the 2003 Fair and Accurate Credit Transactions Act bears testament to the ways in which credit and debt relations are the quintessential condition and materialization of social relations in neoliberal economies (Lazzarato, 2011), as well as the ways in which credit scoring is part of the ‘essential equipment’ of the credit consumer (Langley, 2014: 421). But the social consequences of the crisis also bear testament to how the false promises in President Bush’s proclamation above are not new to Americans, particularly those who have historically been treated unequally along the lines of class, gender and race (Dymski, 2009; Gandy, 2009: 106, 111–115; Rugh and Massey, 2010; Soederberg, 2010). They are certainly not new to South Africans who, depending on group assignment, have benefited and/or suffered from state-implemented categorization for more than 150 years. Moreover, inherited forms of social classification persist in South Africa 21 years since the attainment of political democracy. Their uses on both sides of 1994 as state-legislated strategies of population management inform the characterization of individuals and populations as risks. While framed by the National Party government as a strategy of ‘separate development’, known ubiquitously by its enforcers and opponents as ‘apartheid’, inherited forms of social classification are repackaged by the African National Congress (ANC) government in the terms of ‘social transformation’. The importance of ethnic, gendered and racial classification, we are told by the ANC (2012), is to trace the extent of social change away from historical inequalities. Despite four presidents steering the new ship of political democracy over the past two decades, the message from the ANC of ‘social transformation’ and ‘non-discrimination’ has largely been consistent in terms of the promise of the benefit to be derived from credit financing (Mandela, 1994; Mbeki, 2002; Zuma, 2011). ³

South Africa’s Department of Trade and Industry (2003: 30) frames the importance of credit financing, within a narrative of affirmative action known as Black Economic Empowerment (BEE), as improving the living standards of ‘Black’ consumers. More particularly, credit financing and credit scoring are legislated as promoting ‘non-discrimination’ and protecting against ‘unfair discrimination’, respectively (Government Gazette, 2006: 2, 90). Credit scoring, the dominant way in which the terms of credit financing are assessed by financial institutions, refers to statistically based predictive assessments of how ‘risky’ a consumer is likely to be in meeting the conditions of loan repayment. We should not separate this from the overarching policy of ‘social transformation’ that describes a process of social change away from historically grounded and apartheid-promoted forms of inequality and underdevelopment, towards individual empowerment, equality, freedom and opportunity on the basis of ‘class, gender and race’ (ANC, 2012: 2). Therefore, when we consider credit scoring through the lens of ‘social transformation’, ⁴ we see how it is intended to focus on historical redress of social inequality. Like many Americans, South Africans who were once the subjects of social exclusion are now presented with the possibilities that are offered by social inclusion, or, more broadly, social transformation. This shift has contributed to South Africa being home to the African continent’s most sophisticated banking and credit systems, with approximately 18 million credit-active consumers, of which more than half are women (Transunion, 2014). ⁵ Despite the sophistication of the systems, the country is in the midst of a sub-prime crisis, which suggests the failure of credit legislation, policy and scoring to facilitate social transformation as envisioned by the ANC. This raises questions as to the extent to which credit serves the interests of historical redress in general, and of social transformation in particular. Is, as state legislation infers, the constitution of credit risk apolitical, inclusive, neutral and objective? Have credit legislation and policy since 1994 reconfigured relations of power from their historical formations?

In addressing such questions, it is helpful to consider Geoffrey Bowker and Susan Star’s (1999: 224) remark that ‘the South African case [of racial classification] relates directly to all questions of information systems design where categories are attached to people’. Against this backdrop, I argue that credit scoring, although not explicitly designed as a security device, enacts (in)security in South Africa. By paying careful attention to the state-implemented social categories that the ANC government inherited in 1994, we see how the very same categories that were once used for social exclusion were not dismantled with the dawn of political democracy. Instead, they are framed as measures of social transformation. By analysing social transformation through the lens of credit legislation, policy and scoring, I argue that the latter, in particular, influences, shapes and even reproduces historical forms of social categories in the state’s attempt to create populations as risks. Social categories, typically gendered and racialized by the National Party, were used as a means of controlling, influencing and managing South Africans. Ironically, the same social categories are now a defining feature of the ANC, which uses them to track, trace and transform historical conditions of social existence that continue to shape South Africans’ life chances. In fact, through the lens adopted in this article, there is seemingly a distinction between the social categories used for reinforcing social disparities and those used for observing them. However, I argue that this distinction needs to be reconsidered, because the categories of observation that are highlighted actually contribute to the reproduction of historical social categories. By examining credit scoring in this way, we see how it is framed within the narrative of social transformation, yet is more accurately a socio-technical device used to manage populations as risks.

This management is performed largely through processes of social sorting. To illustrate how social sorting is enacted by credit scoring, and how the latter reproduces historical social categories, this article is divided into four main sections. The first of these briefly explains the theoretical framework that dovetails the scholarly contributions of the sociologists David Lyon and Harold Wolpe. The subsequent section clarifies the term ‘socio-technical device’ that is developed in this article and looks briefly at the relationship between such devices and historical social categories of populations as risks. It is then shown how many of these categories have become normalized to their disadvantaged subjects. The third of these four main sections links credit policy and social transformation, and accounts for how the Minister of Trade and Industry, Alec Erwin, sought to regulate particular kinds of micro-lending that otherwise escaped the credit scoring of registered bureaus. Against the backdrop of the National Credit Act (2005), the fourth main section hones in on credit scoring as a socio-technical device that, through the creation of populations as risks, reinforces historical social categories.

Sorting through the contributions of David Lyon and Harold Wolpe ⁶

At the level of theory, I locate the broader argument within a Wolpean framework and detail the specific significance of social sorting to the enactment of (in)security in South Africa. Wolpe (1988) demonstrates that, in that country, capitalism and the racialization of class are contingently, not necessarily, related. He shows this by intersecting racist ideology with political practice and the capitalist mode of production to demonstrate how they connect to the reproduction of cheap ‘Black’ labour (Wolpe, 1972). The latter was regulated through policies of segregation and apartheid. However, the last quarter of the 20th century saw a decline in the need for such labour by the capitalist system, together with the growing cost of ‘Black’ opposition that threatened the state’s resources in maintaining political stability. Such an analysis of labour-force contradictions of capitalism invokes discontinuities of policies, while a growing ‘black bourgeoisie’ signals ideological continuity (Wolpe, 1988: 49). Accordingly, we see a temporal, not structural, relationship between apartheid policy and capitalism. By furthering Wolpe’s view, we find in credit legislation, policy and scoring grounds for the argument that a Wolpean framework may be extended in the articulation of capitalism and the social categories of apartheid policy as not only historically, but contemporarily, contingent.

On this basis, I contend that Wolpe’s analysis has currency for contemporary understandings of capitalism, class and labour as they relate to credit legislation, policy and scoring. In this perspective, we see that capitalist processes rejuvenate themselves across South Africa’s transition from inwardly projected apartheid under the National Party to US-backed neoliberalism under the ANC (Bond, 2003, 2004). While the political legitimacy of neoliberalism has come into question, particularly following the most recent global financial crisis (Peck et al., 2009), this must not overshadow its channelling of wealth from an impoverished majority to a minority of political and economic actors (Harvey, 2007).

Lyon’s popularization of ‘social sorting’ aids our understanding of this channelling by explicating the ways in which populations as risks are managed. He uses this term to refer to a ‘focus on the social and economic categories and the computer codes by which personal data is organized with a view to influencing and managing people and populations’ (Lyon, 2003: 2). When we consider the social repercussions of social sorting, we see how it may, for example, (re)inform gendered divisions in insurance and profile individuals in the name of financial security.

For gender, biases inform the blanket perceptions of male drivers by a South African insurance company, 1st for Women, that only serves women clients. The company asserts that ‘statistically, women are lower insurance risks than their male counterparts and the cost to repair a vehicle crashed by women is on average, lower than the cost of damage caused by men’ (1st for Women, 2014). For security, Anthony Amicelle (2011: 268) makes clear, with respect to profiling techniques used in the tracing of financial flows, that it is not the framing of social activities as risks that necessarily brings them under the control of security agents, data analysts or other social actors. Rather, it is that the perception is created of risks as necessarily manageable. Ana Canhoto and James Backhouse (2007) extend this view by arguing that segmented groups, whose interpreted characteristics align them with the activities of money laundering, may unwittingly be discriminated against by financial institutions. Similarly, Marieke de Goede’s (2012) concept of ‘speculative security’ explains how surveillance mechanisms in finance are employed in a temporally reoriented fashion that shifts from punishing past crimes to future terrorism. She frames this within the narrative of the War on Terror and demonstrates the social consequences, for example, on Muslim immigrants to the USA that fund organizations such as Islamic charities. Through data mining and profiling of financial transactions, such individuals are profiled and put on a ‘watch list’. In the same way that this is a strategy to confront terrorism by laying out punishment based on a pre-emption of intentions, credit scoring predicts future consumer behaviour to determine levels of punishment (or rates of interest) accordingly. While ‘financial terrorism’ relies on social sorting to exclude individuals on the grounds of providing protections to territory and a group of ‘insiders’, the kind of social sorting that I deal with in the remainder of this article is implemented by the South African state on the basis of social rehabilitation and inclusion.

When we consider such inclusion in the framing of ‘social sorting as social transformation’, ⁷ we see that inclusion not only captures the ways in which a policy narrative justifies social sorting by virtue of social transformation, but also enables critical examination of such a narrative that mutually constitutes these terms. By extending Wolpe, as stated above, we position our focus squarely on social sorting and the similarities between the social categories of rule on both sides of 1994. These categories are framed in the ‘new’ South Africa as indicators of social transformation. In reality, we see that social transformation not only influences, shapes and reproduces historical social categories, but also serves to create and maintain populations as risks in the governance of South Africans. Three recent examples collectively illustrate the intersection of Lyon and Wolpe on how historical social categories continue to inform the state’s identification of its citizens.

First, a Statistics South Africa (2009) study identifies ‘changes in the standard of living in South Africa overall and for the four main population groups’, which are sorted as ‘African’, ‘Coloured’, ‘Asian’ and ‘White’. It observes that ‘there is no evidence of deterioration in the material standard of living of White households’, as if to suggest that the Prohibition of Mixed Marriages Act (1949) and Immorality Amendment Act (1950) are still in effect (Statistics South Africa, 2009: 1, emphasis added). Both of these Acts sought to regulate the personal and sexual relations between individuals on the basis of race. Such a study is problematic not only for defining a household by race, but also for giving continued support to historical forms of sorting that allegedly provide a basis for monitoring social change through class mobility (Wolpe, 1988: 13).

Second, in 2008, Chinese South Africans won a court case in favour of their classification as ‘Black’ (Park, 2011). Under the broader scope of social transformation, this was a victory that legalized their benefit from BEE. Over the last century, the groupings of Chinese South Africans have been transient, with state-imposed social classifications shifting between ‘Non-white’, ‘Coloured’ and (conditional) ‘White’. This makes apparent the political contestation in the categorization of individuals and populations.

Third, bearing an eerie similarity to the numerical colour-coded charts used under apartheid is the continued usage by at least one of the ‘big four’ banks of the codes ‘0-1-2-3-4’ to racially identify its clients. These refer to the respective categories of ‘White’, ‘Indian/Asian’, ‘Coloured’, ‘Black’ and ‘Other’. On an anthropological visit to this bank, I observed that my default classification on file, since 2002, was ‘1’. I was informed that such classification was required in order to trace racial changes in wealth. Yet, it was with relative ease that a bank clerk allowed me to ‘update’ my racial classification to 0. No questions asked. Not only is the numerical sequence of the categories symbolic of an historically structured system of privilege, but there is an apparent ease with which one may move across categories. While this ease is indicative of a focus on ‘self-identification’, it also poses further questions about the accuracy of studies, such as Statistics South Africa’s quoted above.

In these examples, we see a concern with racial concepts, raised by Wolpe (1972: 428–429) over 40 years ago, that are presently framed at the centre of social transformation. Not only does this concern favour the social construction of racial classification as a definitive method of analysis, but it also reflects the conditions through which racial ideology becomes the dynamic force of South African political economy. By relying on economic and social categories, from which stereotypes and prejudices are often derived, social sorting contributes to subjectivities about populations and how they should be treated on the basis of reward, suspicion or even guilt. These forms of social sorting are often based on assigned group membership that is considered more telling of individual tendencies and behaviours than individuals’ actual actions, activities and social contexts (Gandy, 1993: 18).

Socio-technical devices and the social sorting of populations as risks

A focus on populations as risks as opposed to biopolitical understandings of populations at risk is not intended to preclude an articulation of the ‘disciplinary’ and ‘regulatory’ effects of managing individuals and populations (Foucault, 2004: 252–253). Rather, the objective is to shift focus from conditions of risk to categories of risk. While the former feature ‘systemic characteristics’ of the social environment (such as states of being at risk of disease, death or environmental danger) (Aradau et al., 2008: 148), the latter sharpen the focus on the differential assignment of individuals and groups on the basis of how risky they are deemed by those who perform (or automate) social sorting. In this way, predictive models that identify consumers as risky place them ‘at higher risk because of their identification as risks’ (Gandy, 2009: 133, emphasis added). ⁸ While Michel Foucault’s theory of biopolitical government, or the concept of neoliberal governmentality, may be of relevance here, it is equally important to carefully consider the application of exogenous theoretical frameworks that show little, if any, direct interest in ‘the colonial’ social formation in general (Legg, 2007: 265–269), or in the particular historiographies of South(ern) Africa in particular (Breckenridge, 2009: 41).

By considering consumers as risks, it is apparent that the social categories of credit scoring reify consumer attributes (Marron, 2007). The higher the risk of lending to a consumer, based on analytically determined categories, the higher the chance of predicted default that is naturalized as individual responsibility (Marron, 2007). Within this context, a focus on credit scoring as a ‘socio-technical device’ allows us to question its framing in legislation and policy as an apolitical tool. In fact, the task of politicizing credit scoring as a device is contingent upon separating it from its normalized institutionalization. As important is the need to examine the spatiality – or ‘lendingscapes’ – of credit scoring (Burton, 2012). This not only brings attention to differences in credit conditions, as Alya Guseva and Akos Rona-Tas (2001) demonstrate with respect to Russia and the US, but begs questions about the spread of ‘Western scoring methods’ to non-Western contexts (Burton, 2012). While Andrew Leyshon and Nigel Thrift (1999) frame credit scoring as a tangible device used to embody knowledge about consumers in retail finance, Rona-Tas and Stefanie Hiss (2010: 148) offer a more conceptual alternative in their proposition of credit scoring as a ‘sanctioning device’ that marks a shift from complex prediction to transparent rules in loan contracts. The view taken in this article is that of Martha Poon (2009), whose conceptual framing of credit scoring grants her the vantage point from which to see the credit scorecard as a socially positioned device that constitutes markets. This allows us to examine credit scoring as a socio-technical device that is positioned at an important intersection between economic markets, consumer experience and political ideology.

To move beyond an apolitical framing of credit scoring that positions the constitution of markets as dependent on pre-existing market forces as opposed to the activities of people requires an appreciation of risks as constructed, not inherent. François Ewald (1990: 142) notes that ‘nothing in itself is a risk – risks have no real existence’; rather, they are reorganizations of reality. If we consider this reorganization as an effort to quantify losses and probabilities (O’Malley, 2004), then we see how it may be used to institutionalize individualization (Beck and Beck-Gernsheim, 2002). We also see how risks are not necessarily eliminated, but consciously embraced, by their enforcers (Aradau et al., 2008: 148). Michael Dillon (2008: 319) demonstrates that the pursuit of profit and the securing of governance may be mutually constituted, and that the operationalization of risk is at the centre of this relationship.

The ways in which populations in South Africa were created and managed as risks is, in many respects, what separates the country’s case of population management from practices in many other parts of the world. Indeed, its people are no strangers to social classification, matching, clustering and segmentation, which they have known for more than 150 years. Bureaucratic obsessions with race in South Africa through processes of social sorting include: the 1858 Grondwet (Constitution) of the Zuid-Afrikaansche Republiek, which offers classifications of gekleurden (Coloureds) and blankes (Whites) (Suzman, 1960: 339–340); the (Mahatma Gandhi-motivated) fingerprinting of Indians in Natal and the Transvaal in the early 20th century (Breckenridge, 2011); and experimental technologies of ‘scientific’ classification under apartheid for the racial sorting of all citizens in the second half of that century (Hamblett, 2014).

Wolpe (1972: 432–433) argues that a contradiction of the National Party’s project of apartheid was that it relied heavily on the reproduction of cheap labour from Bantustans (Homelands) while the social conditions for the reproduction of that labour disintegrated. The social sorting of individuals that was used, for example, to regulate the flow of waged labourers depended on a number of socio-technical devices that the National Party administered to classify people into groups that could be managed differently. Pencils were used to conduct ‘scientific tests’ of racial difference by comparing their movement through different hair types (Posel, 2001: 59). Perhaps most well known are the bewysboeke (books of proof) that had to be carried by those classified as ‘Black’ in order to regulate their movement in designated ‘White’ areas. The racialization of class, socially shaped with the influence of such technologies, was a key component in the enforcement of minority privilege and social disadvantage for most South Africans. The classification of people by race informed divisions of labour that were used for the maximization of profit (Keegan, 1983; see also Foucault, 2004: 258). Equally, the racialization and regulation of social and physical spaces were used to the ends of, for example, racial conformity and social discipline (Posel, 2010).

We see from these examples that social sorting sits at the centre of practices of unequal rights (Wolpe, 1988; Posel, 2001) predominantly based on one’s racial classification. While a plethora of research captures the ways in which waves of resistance confronted state practices of inequality, less critical attention is granted to the normalization of social sorting by those who suffered its greatest social consequences (Posel, 2001). On the one hand, we see this bifurcation in a rejection of National Party rule both domestically and internationally, as well as in representations of political agency, non-compliance and political intent that undermine such rule. On the other hand, as much as social constructions of race under the National Party were maintained by a heavily surveillant state, they were also embraced by social groups, seemingly with Fanonian (1963) inspiration, as modes of unity. 1970 marked a rejection by members of the ANC of the label ‘Non-white’ in favour of the preferred ‘Black’ (Gumede, 2007: 28). Other examples of self-appropriation of group identities include: the South African Indian Congress (SAIC), the Coloured People’s Congress (CPC), and Stephen Biko’s Black Consciousness Movement.

Therefore, political and social formations were largely shaped by social realities that racialized and inhibited the freedom of social and political association. Even when writing in the late 1980s, it was clear to Ratnamala Singh and Shahid Vawda (1988: 10) not only that state-imposed racial groupings indicated the terrain of apartheid influence that racially segmented, but also that ethnically constituted political self-organization risks the reproduction and reinforcement of assumed ethnicities. This offers expression to the ways in which the treatment of populations as risks was not eradicated with the exit of the National Party from political office, but transferred to the ANC when it assumed the same political position. This transfer is framed by the ANC as social transformation because of the effects of historical social sorting that contribute to self-identifying social categories (Posel, 2001). With such identity constructions designed into everyday life, their ubiquity unwittingly forms part of epistemological understandings of social formation. It is arguably through such a mechanism that the social sorting of South Africans since 1994 is presented by the state as empowering and bearing testament to the nature of structural social changes that address historical social inequalities.

Legal and social enactments of credit policy as (in)security in South Africa

During a personal interview with a credit scoring analyst from one of South Africa’s largest credit bureaus, it was put to me that ‘[credit] scoring is starting to get a bit over-rated’. While this view suggests that critical attention to credit scoring is overestimated, I demonstrate in the remainder of this article that the opposite is true. Particularly against the backdrop of social sorting as historically contingent in South Africa, credit scoring is a salient, yet under-rated, socio-technical device in the creation and management of populations as risks.

While credit scoring is not typically included in literature on finance and security, the case to do so is articulated here by demonstrating the role of credit scoring in enacting (in)security in the political economy of finance. The transition to a ‘democratic’ South Africa in the early 1990s was a period of economic and political compromise, tension and resistance. Politically, the presidency of F. W. de Klerk (1989–1993) exacerbated conflicts in the ruling National Party, particularly evident in the opposition to him from conservatives and the South African police (Glad and Blanton, 1997: 572). Economically, capitalist benefactors of the apartheid regime feared that the decline of the National Party was jeopardizing their financial control of the economy (Bond, 2003: ix, 272; 2000; Wolpe, 1992: 3). In fact, the benefactors of the credit system of the 1970s and 1980s that primarily served the interests of the ‘white middle class’ feared that that system was coming to an end (Department of Trade and Industry, 2003: 8). As a ‘defensive action’ of the regime-in-decline to hold on to economic and political power, the National Party successfully compromised with the ANC in the transition to a joint Government of National Unity (Wolpe, 1992: 3). While new economic policies were put in place, as South Africa shifted from a mostly state-led market arrangement to a globalized free-market economy, these policies did not signal the dismantling of dominant economic and social forces as much as they did an adaptation to such forces (Wolpe, 1992: 3).

Among the values of political democracy that dawned in South Africa with the election of President Nelson Mandela was the embrace of the market conditions that presently shape the geopolitical trajectory of the country. In fact, market liberalization brought with it credit liberalization (James, 2012: 24) in ways that substantiate social transformation as a project of social inclusion, empowerment and individual destiny (ANC, 2012). For example, South Africa’s new economic direction was largely shaped by the ANC’s election manifesto, the Reconstruction and Development Programme (RDP) and the ongoing Growth, Employment and Redistribution (GEAR) policies (Bond, 2000, 2004; Desai, 2004; Habib and Padayachee, 2000). The GEAR policies look favourably upon credit as a remedial tool for apartheid-promoted inequalities (Department of Finance, 1996). These primarily take the forms of credit-guarantee facilities for all South Africans and, reflective of a neoliberal and global policy push of the General Agreement on Tariffs and Trade, the relaxing of barriers for foreign investors seeking access to domestic credit. The RDP makes explicit mention of the kind of remedial action expected to be derived from credit financing. For example, Section 4.7.3 of the RDP states:

The democratic government must, in consultation with financial institutions, establish prudent non-discriminatory lending criteria, especially in respect of creditworthiness and collateral; reform the laws on women and banking to ensure equality; forbid blanket bans on mortgage bonds to specific communities (‘redlining’); require banks to give their reasons when turning down a loan application; establish community liaison boards; develop simpler forms for contracts and applications, and create an environment which reduces the risk profile of lending to small black-owned enterprises and requires banks to lend a rising share of their assets to small, black-owned enterprise. The law must also require that financial institutions disclose their loans by race and gender; their assets and liabilities by subregion and sector; their staff by race and gender; the location of their branches and defaults by neighbourhood. (RDP, 1994, emphasis added)

The centrality of access to credit in this excerpt is made clear in its identification as a key component of social transformation in the RDP. Consequently, this may be read as a seemingly anti- or non-discriminatory programme in response to charges of historical discrimination against persons and communities. However, one must recall from the previous sections of this article that evaluating social transformation in the terms of inherited classifications validates historical social categories that continue to be constructed. This clash between remedial rhetoric of utopianism and the practicalities of social rebuilding normalizes the management of populations as risks. This is because the collection of data on gender and race is framed as serving the interests of social equality while also justifying in the mechanism of credit risk the presence of categories that promote social inequalities. In this way, efforts to rearticulate gender and race as depoliticized constructs that escape their historical experiences actually normalize their inclusion in the reconstruction of populations as risks (Posel, 2001).

The Usury Act Exemption (1992) aimed to expand credit activities in the formal sector, particularly by including those low-income (especially rural and poor) households otherwise deemed too risky to gain access to credit in the formal sector. South Africa’s new free-market environment in the early 1990s permitted the introduction of this Act as one of the first changes implemented by the political coalition. Aware of the contributions made by micro-lenders to low-income households in the informal sector of the economy, the implementers of the Act sought to regulate such economic activity by enabling micro-lenders to become legally registered enterprises (Daniels, 2004: 837). However, the free-market conditions that informed Minister of Trade and Industry Alec Erwin’s establishment of a credit-friendly environment meant that micro-lenders were ‘aggressive’ in their credit marketing strategies, charging unregulated – often high – rates of interest on loans to micro-, small- and medium-sized enterprises (James, 2012: 24). Alternatively referred to as loan sharks, they were equally aggressive in their collection techniques, in that they often kept the bank cards, identity books, payslips and other personal details of those who borrowed from them (Hlongwa, 1998).

Free-market conditions also enabled the Act to legalize high rates of interest that loan sharks were already charging those consumers deemed by the formal sector as ‘high risk’ (James, 2012: 24). In fact, loan sharking expanded to a R10 billion business by 1998, a 280% increase over one year (Hlongwa, 1998), and was reducing political control of the credit industry. In light of this, Erwin proposed revisions to the Usury Act Exemption (1992) to place a ceiling on the financial charges implemented by micro-lenders and also force them to register with an approved regulatory authority. While signalling to consumers concern over abusive credit practices, this can be seen as a political effort to redirect the funnelling of credit to consumers from the formal sector. This was intended not only to increase the accumulation of corporate tax in the hands of the state, but also to grant greater control to policymakers who otherwise could do little to regulate the devolution of lending practices. Despite Erwin’s efforts, the courts did not welcome the proposal of imposing an interest ceiling (Erwin was in search of a ceiling ten times that of the prime rate) on the basis that it would contribute to driving micro-lenders underground (Daniels, 2004: 847; Whittaker, 2008: 570). With minimal access to credit in the formal sector, and excessive rates of interest from loan sharks, low-income individuals found it difficult to demonstrate their abilities to manage credit, and in many cases to even begin establishing credit histories. For many South Africans, this had the consequence of extending similar perceptions of them by commercial banks as ‘too risky’ that were initially motivated (although not legally enforced) under National Party rule (Hull, 2012: 168). By regulating, and not banning, micro-lenders, the newly formed Microfinance Regulatory Commission (MFRC) hoped to encourage them to not go underground, where they would charge even higher interest rates. Notably, this illustrates how a system of regulation was formed to establish systems of rights, including legitimacy. With the proposal of interest ceilings rejected, it was decided that usury laws would encourage micro-lenders to engage in ethical and transparent behaviour. However, with little consumer protection from the Usury Act Exemption (1992), underground activity persisted.

While the unregulated supply of micro-credit was deemed a practice of ‘reckless lending’ (James, 2012: 21), there were strong calls from the South African Reserve Bank (SARB) for the deregulation of credit bureaus. In fact, with at least six credit bureaus operating by 1999, the threat of oligopolies led the SARB to express the importance of deregulation. This was largely due to concerns with the Joint Bank Credit Bureau that managed credit applications exclusively through the four main banks, and was seen as operating a proprietary database that denied access to non-members. This was perceived by the SARB as a disadvantage to new and smaller banks, because it was a hindrance to the availability and quality of consumer credit information. Largely favouring the ‘modern banking trends’ of, in part, unregulated international credit bureaus (in the US and Canada in particular), impetus was given to the entrance of such new and smaller institutions to promote accurate consumer information that most banks and retailers were otherwise excluded from (SARB, 1994: 46; 1999: 30, 32).

With decade-long parliamentary debate on the conditions of credit practices, there were calls for more comprehensive management with legislative enforcement, to promote responsible borrowing behaviour and to provide consumers with credit protections. This finally led to the passing of the National Credit Act (2005), which we examine in the following section.

The National Credit Act (2005) and credit scoring as a socio-technical device

The National Credit Act replaced the Usury Act Exemption (1992) and Credit Agreements Act (1980). It came into effect on 1 June 2006 and outlines access to credit as an individual ‘right’ (Government Gazette, 2006: 88). In this way, the National Credit Act (2005) individualizes the management of credit and the responsibility of debt (James, 2012: 27–28; Schulze, 1997: 24). Given the pertinence of social transformation to the Act, we see from this individualization in credit law how responsibility for historical forms of social inequality shifts from the state to individuals. Through micro-lending in particular, previously unbanked populations are brought into the formal financial sector, which allows for the inclusion of a greater population as ‘manageable’ risks. The extension of credit is not always at the forefront of risk analysis, as we saw in Phuthaditjhaba (in the province of the Free State), where pre-loaded debit cards were issued to 5,000 consumers. Significantly, this was a world-first pilot project spearheaded by Mastercard and Capitec Bank (Merten, 2005). This example is significant for two main reasons. First, Capitec benefited from start-up capital provided by the United States Agency for International Development (USAID) to the amount of R50 million. This flow of funds must be seen as contrary to the claim three years earlier in a confidential US government report about USAID’s ‘meddling’ in South Africa’s policymaking (cited in Bond, 2003: 84). Second, the banking of poor, formerly unbanked, populations provides financial institutions, such as Capitec, with personal consumer information that can be used in future credit checks. ⁹ In fact, Capitec has mushroomed into one of South Africa’s largest micro-credit providers among the commercial banks. Particularly because credit checks are required by law prior to approving the lending of funds, we see how the extension of credit to ‘risky’ groups cannot easily be separated from Capitec’s CEO, Riaan Stassen, earning remuneration packages of R10.5 million and R10.8 million in 2012 and 2013, respectively (Fin24, 2013). It is the construction of consumers as risks that, some argue, significantly contributes to Stassen’s personal wealth, and his ability to take early retirement in 2013 (Bateman, 2012).

A managing director from Nedbank (one of the four major commercial banks) commented on the expansion of credit to low-income areas in South Africa: ‘It is a very attractive market for us [commercial banks] … there is a fortune to be made at the bottom of the pyramid’ (cited in Bank Marketing International, 2006). Unsecured personal loans are one kind of credit financing that speak to this perceived ‘attractiveness’, especially since approximately 68% of revenue earned from these loans is derived from interest rates alone (not to mention from additional fees and penalties) (Compliance and Risk Resources, 2012: 63). Similar corporate principles emerged from a one-day seminar, ‘Marketing to the Low-Income Consumer’, held in Sandton, Johannesburg, in 2012. It was organized on the premise that low-income consumers are a ‘brand’ that can be targeted in specific income-generating ways (Knowledge Resources, 2012).

The National Credit Act (2005) makes specific mention of previously disadvantaged South Africans, such as those described above, who are a significant source of corporate profit. In so doing, the Act promotes the assumed benefit for consumers to be derived from credit financing (Government Gazette, 2006: 18). This ideal, framed more broadly as improving access to credit for the poor, was a campaign initiative of the South African Communist Party in support of the enactment of the National Credit Act (2005) (Blade Nzimande, cited in Tabane, 2011). The Act’s stated purpose is ‘to promote a fair and non-discriminatory marketplace for access to consumer credit’ (Government Gazette, 2006: 2). It also stipulates that a credit provider ‘must not unfairly discriminate directly or indirectly against any natural person, juristic person or association of persons on one or more grounds set out in section 9(3) of the Constitution’ (Government Gazette, 2006: 89). That specific section of the Constitution of South Africa details discrimination, and by inference an expression of unfairness, on the grounds of ‘race, gender, sex, pregnancy, marital status, ethnic or social origin, colour, sexual orientation, age, disability, religion, conscience, belief, culture, language and birth’ (Government Gazette, 1996: 1247). Notably, if credit scoring, by extension, contributes to the promotion of discrimination and unfairness on these grounds, then it fails in serving as a tool of social transformation. But it also seems that credit scoring may contribute to the highlighting of gendered financial burdens. This is evident when we consider two main points. On the one hand, women comprise the majority of credit-active consumers (Transunion, 2014). On the other hand, Statistics South Africa (2012: 74) estimates that households headed by unemployed women remained in the region of 43–48% between 2002 and 2011. Although there is no direct correlation offered by Statistics South Africa between the two findings, my preliminary analysis of open-ended interviews with unemployed Durban women in possession of bank loans suggests a notable link. ¹⁰

We see in the National Credit Act (2005), as we did earlier in this article, the ways in which the ANC upholds many of the historical social categories that were legally enforced and socially experienced. Even with the ANC’s intention of allegedly using those historical categories as a basis for tracking social improvements, they do not escape being framed in the language of social sorting. While the right of access to credit may be emancipatory, the context in which it is framed, promoted and protected, by implication, also provides justification for its control and management. As we see above, these functions are often fulfilled under the auspices of social transformation by authorities such as commercial banks and the state (for example, the South African Police, Statistics South Africa and other state affiliates and departments). Indeed, the promotion and protection of credit rights demand numerous methods of very visible surveillance – perhaps reminiscent of National Party state surveillance – which arguably (and ironically) undermine the notions of freedom that such surveillance hypothetically protects. Enumerating certain constitutional rights, therefore, serves as a precondition for managing populations on the basis of calculable risk analysis.

This takes shape in the National Credit Act (2005), which offers legislative weight to the functions of credit bureaus and the normalization of risk analysis as calibrated in their credit assessment techniques (Bank Marketing International, 2006; Whittaker, 2008: 276):

A credit provider may determine for itself any scoring or other evaluative mechanism or model to be used in managing, underwriting and pricing credit risk, provided that any such mechanism or model is not founded or structured upon a statistical or other analysis in which the basis of risk categorization, differentiation or assessment is a ground of unfair discrimination prohibited in section 9(3) of the Constitution. Government Gazette (2006: 90)

What this overlooks is that even if the ‘basis of risk categorization’ is based on intentions of ‘unfair discrimination’, it does not preclude social experiences with credit that are unfairly discriminatory. Research, for example, on the scoring methods of the US- and Irish-based credit bureaus Experian and Transunion, respectively, draws links between credit scoring, consumer vulnerability and consumer profiling (Mierzwinski and Chester, 2013). These credit bureaus perform credit assessments of millions of consumers in 58 countries, including South Africa (Experian, 2011; Transunion, 2011). They use FICO (known as Fair, Isaac and Company until 2003) scoring models for their credit risk assessments. The fact that Experian’s Delphi and Transunion’s Empirica credit scores, together with the scoring methodologies of the other 12 registered credit bureaus in the country (NCR, 2014a), are marketed as unique and competing products illustrates the scientific imperfection of credit risk. Therefore, algorithmically generated sketches of consumers, defined along axes ranging from ‘low risk’ to ‘high risk’ (typically, the higher the score, the lower the interest rate that is attached to a loan), compete for ‘accuracy’ across competing scoring models. As noted above, imposed racial classification is not tolerant of a person’s socio-cultural identity but instead deflates the quality of life in favour of imposing labels from finite categories. We see a continuation of this logic, with legislative weight, under the cloak of avoiding ‘unfair discrimination’.

But there is supposed to be recourse for the consumer with the National Credit Regulator (NCR) (conceived in the National Credit Act (2005)). The NCR oversees all credit-related transactions in South Africa, with an official mandate to ‘serve the needs of (i) historically disadvantaged persons; (ii) low income persons and communities; and (iii) remote, isolated or low density populations and communities’ (Government Gazette, 2006: 46). Despite this responsibility, the National Credit Act (2005) offers few guidelines as to how the NCR is to fulfil its functions. The Act is also tight-lipped on how the NCR is to assess whether industry-standard procedures are being followed by various commercially run financial institutions, credit bureaus, debt collectors and other market participants. What the NCR does offer the consumer is assurance that credit bureaus ‘may not list information that may be discriminatory’; yet, by implication, it does not hinder the collection of information pertaining to ‘race, sexuality, political affiliation, medical status, religion or membership with a trade union’ for purposes such as benefit, influence or management (NCR, 2014b, emphasis added).

Indicative of the value of such information to credit bureaus was the industry uproar that followed the introduction of the National Credit Amendment Bill (2014) on the grounds that it would interfere with credit providers’ assessments of consumer risk (Ensor, 2013). In part, the Bill details the automatic removal from credit-bureau records of blacklisted consumers who have settled their accounts (Visser, 2014). The credit amnesty that came into effect on 1 April 2014 gave credit bureaus two months to update their records accordingly. It is unclear whether this is reflective of certain arguments in civil society that particular kinds of consumer information ought not to be included in credit risk assessment. However, it makes evident that the removal of consumer data from credit-bureau records, following the settling of personal accounts, will affect the design of credit scoring that is heavily reliant on historical data (Credit scoring analyst from a registered credit bureau, personal interview, 2013). When considering the role of credit scoring in governing (in)security, we should also keep in mind that the credit amnesty was introduced just five weeks before South Africa’s general election that saw Jacob Zuma’s presidential re-election. This was branded by opposition parties as a move of ‘populist vote-buying’ (eNCA, 2014). With just under 50% of credit-active South Africans in arrears, the ANC is all too familiar with the potential of financially disgruntled and protest-intensive groups. It learned this, most recently, from xenophobic attacks in Durban in April 2015, and from the Marikana Massacre of 2012. The latter drew attention to wildcat strikers in search of higher pay and ended with fatal consequences for a number of mineworkers. In the same way that, for Wolpe, the National Party’s project of ‘separate development’ was Sisyphean (cited in Freund, 2013: 496) in being beset with internal contradictions of labour and capital (Wolpe, 1972), we find such characteristics also emerging from the Marikana crisis.

Reminiscent of labour-force contradictions that affected waged workers in the National Party’s project of apartheid is the social positioning of Marikana miners as financially strained, perhaps even stuck in a ‘debt trap’ (Alexander, 2013: 607). It seems that even improved social conditions of labour under the ANC (as compared to under the National Party) (Freund, 2013: 512–513) are inadequate to overcome the social reproduction of poverty that partially triggered the Marikana crisis. Despite social improvements for mineworkers under the ANC (such as higher real wages, bargaining contracts with worker unions that provide medical insurance and a choice of provident fund or routine pension payments, a housing allowance, and the payout of bonuses when mineral prices increase), the Marikana Massacre raises questions about social strains on workers for which contemporary labour conditions are unable to compensate. Elements of these strains are combinations of indebtedness (such as miners invoiced to pay up to 15 times the value of original loans), increasing prices for electricity and water, short-term labour contracts, and adversarial treatment by managers often on racial grounds (Alexander, 2013: 607–608; Freund, 2013: 513–514; Malcolm Rees, cited in Bond, 2013: 582–583). As Bill Freund (2013: 514) notes,

having to dispense with pension and health benefits for immediate cash as well as falling into debt are the fate of the apparently better-off, and the shift to contract employment is thus sometimes actually welcomed, so strong is the compelling short-term argument.

It is this short-term argument that we find inherent in the nature of ‘easy’ and ‘quick’ access to micro-credit. At a ‘macro’ level, the Massacre demonstrates the interplay of local and global powers of policing (Bond, 2013); upon downgrading the state’s credit status from A3 to BAA1, the Moody’s rating agency (cited in Derby, 2012) commented on the government’s ‘reduced capacity to handle the current political and economic situation’. This reminds us of South Africa’s neoliberal integration and heeling to global powers. It also leaves for more detailed examination the difficulties with which credit scoring, as a seemingly domestically focused socio-technical device, can be separated from an international and highly politicized agenda.

Conclusion

The nature of social sorting that is evident through credit scoring in South Africa resembles many classical understandings from other parts of the world of how such sorting deals largely with managing levels of advantage, consumption and entitlement. However, one key difference is that while social sorting was once promoted by the National Party in the condition of apartheid, the framing of such sorting by the ANC government reproduces historical social divisions. Using a theoretical framework informed by Lyon and Wolpe, we see that social sorting is enacted by credit scoring. More particularly, social sorting not only reproduces many historical social categories (and forces us to consider these categories as a condition of apartheid under ANC leadership), but also forces us to pay close attention to the strategies that the ANC employs in an effort to present a discontinuity from the past. The point is not simply to state that historical social categories are reproduced, but to critically examine the devices, processes and tactics that are framed as offering breaks from historical social conditions. I developed one part of that story by focusing on credit legislation, policy and scoring. Because this trio is encompassed by the narrative of social transformation, it promises equality, non-discrimination and objectivity. However, when we position credit scoring, in particular, along a continuum of political efforts that seek to manage populations as risks, we see that the social categories of historical sorting are not dissimilar to those that arguably shape the functioning of credit scoring. The legal requirement for credit scoring to be designed and implemented in ways that avoid the social sorting that the ANC inherited creates a direct link between the social categories of historical sorting and the social categories of contemporary sorting. This link is maintained by the ANC’s promotion of the longevity of inherited social categories in the very terms that historically rendered such categories visible. In this way, the narrative of social transformation serves the transfer of management of populations as risks from National Party to ANC leaders more than it does the kind of social change envisioned in the narrative.

Instead of challenging social sorting in action, the ANC does so mostly in word (Bond, 2004). This is consistent with the ‘exploitable contradictions’ of neoliberalism (Harvey, 2007: 42) and allows the government’s support for credit scoring to offer more to the reproduction of historical categories of social inequalities than to the alleviation of such inequalities. It therefore becomes clear that even when the ANC recognizes its shortcomings in shaping a more equitable society, it often does so in the language of inherited failures rather than through a new ontology of social development. Invariably, references to ‘race’, ‘ethnicity’, ‘discrimination’, etc. are used as the motivation for ANC legislation and policies of redress. As a result, the transformation of social conditions is bound by descriptions that are predominantly classist, ethnicized, gendered and racialized. While the ANC’s challenge of social change should not overlook the structural effects of such inheritances, reference to them in inherited terms of social sorting plays significant roles in perpetuating the social effects of discrimination within these confines.

The argument here does not deny the importance of recognizing the social categories within which patterns of discrimination might be observed. Neither does it deny the framing of discrimination without reference to social categories, nor the framing of inequalities independent of social categories or groups (Gandy, 2009). Instead, the argument brings attention to the shortcomings of the ANC’s approach to social transformation that frames social change for the present and future in the terms of the past.

The point is to be vigilant of the historical wounds of social inequality, but also to envision ways forward where even the conceptualizations of that path are free of the inequality that is sought to be overcome (Haraway, 2004). Perhaps a lesson is to be taken from Lawrence Hamilton (2003), who argues that the ANC’s preoccupation with a political philosophy of rights leads to the neglect of what he terms ‘the political philosophy of needs’. This neglect might be due to development strategies that are epistemologically but not ontologically different from historical policies of ‘separate development’. This shortcoming has resulted, for 21 years, in a reproduction, rather than eradication, of historical social categories on the basis of membership to such categories being rights previously curtailed by former leadership.