Concerns about the fairness of predictive practices have been at the forefront of pivotal political debates in recent years, including debates about algorithmic justice, data politics, and predictive policing. Increased technical capacities for data collection and surveillance and the rise of algorithmic decision-making have renewed and intensified questions and conflicts about the role that predictive practices play in governing our social, political, and economic life.

One of the most notable and intractable issues besetting contemporary predictive practices is their relationship to extant unjust social hierarchies. Scholars of contemporary predictive practices in criminal justice, the tech industry, and financial markets have raised concerns that predictive practices reproduce and exacerbate pre-existing inequalities and relations of subordination, especially those based on ascriptive categories of social difference, such as race and gender (Benjamin, 2019; Harcourt, 2007; Noble, 2018; O’Neil, 2016). They have also warned that prediction, insofar as it is seen and represented as a technical, neutral, and objective process, de-politicizes these inequalities, thus entrenching and reifying them (Benjamin, 2019, pp. 2, 33, 160; Noble, 2018, pp. 1, 37).Footnote 1

In distinction to previous debates about the politics of predictive and calculative practices, these debates have centred questions of racial justice.Footnote 2 They have also more clearly distinguished the discrete ways in which predictive practices can wrong with regards to the maintenance and reproduction of racial orders. Recent critiques of the limits of algorithmic fairness paradigms, for example, have made a strong case that accuracy, objectivity, and representation in predictive practices are insufficient regulative ideals in societies marked by deep injustice (Davis et al., 2021; Hoffmann, 2019; West, 2020). Such critiques helpfully move us beyond solely problematizing instances in which predictive practices are distorted by racist ideologies, and provide a basis for critiquing the ‘mere’ reproduction of a deeply unjust racialized economic order. These debates about the ‘racial lives’ of predictive practices have focused on identifying the (re-)articulation of ideas of race in and through predictive practices, the reproduction of racialized inequalities through predictive practices, and the depoliticizing effects of calculative and predictive practices.Footnote 3Critiques of depoliticization have sought to reveal the hidden political or normative commitments that lie beneath the surface of seemingly technical and objective calculative processes, with the explicit aim of making alternative modalities of future-making possible (Joseph, 2014, p. xxii).

While calls for the re-politicization of calculative and predictive processes abound, there has been comparatively little engagement with existing traditions and modes of politicizing prediction, however. In this article, I argue that anti-racist social-theoretic critiques of prediction should also interrogate existing modalities of politicizing predictive practices. This is important in order to orient demands for—and expectations of—the politicization of predictive practices. It raises the question of why certain modalities of contestation are dominant and how the politicization of prediction is shaped by market institutions that continuously discipline contestations of market-based knowledge production. After all, politicizations and contestations of predictive practices do not take place on a blank slate (cf. Boland, 1989, p. 602). They are shaped by pre-existing, hegemonic modalities of contestation and dynamic relations of contestation.Footnote 4 If demands for re-politicization are not to remain politically vacuous, we must examine what modalities of contestation are dominant, how have they become dominant, and how they shape our political imaginary: What political possibilities do they open up? What forms of political imagination and change do they foreclose?

In this article, I want to focus our attention on the politicization of predictive practices by anti-racist social movements. Anti-racist contestations of predictive practices have a long history in the U.S., especially in the form of struggles against the use of ‘race’—that is to say, the use of racial classifications—in the assessment of creditworthiness. Struggles against red-lining and its afterlives have functioned as, and continue to function as, crucial sites for the articulation of anti-racist critiques of predictive power in the U.S. (cf. Krippner, 2017; Marchiel, 2021; Taylor, 2019; Thurston, 2018).Footnote 5 In the following, I examine the paradigmatic example of anti-predatory lending activism in the decade preceding the Great Financial Crisis of 2007/2008 to trace how activists within the anti-predatory lending movement critiqued and politicized predictive practices. Through a close reading of the debates about race, risk, and prediction in the run-up to the subprime crisis, I identify and critically examine the modality of politicization that continues to dominate anti-racist contestations of predictive economic practices in the contemporary U.S.Footnote 6

I argue that actuarial conceptions of fairness-as-justice reverberate in contemporary anti-racist politicizations of prediction. ‘Actuarial conceptions of fairness’ here refers to the normative idea that each individual is responsible for the risk they ‘represent’, and ought to be treated in accordance with ‘their’ risk. The central distributive norm of actuarial fairness can be summed up in the slogan: ‘to each according to their risk’. This conception of fairness, I argue, emerged with the widespread adoption of actuarial calculations of individual risk expectations in the private, for-profit insurance industry. The dominant mode of contesting predictive power, which I will call a ‘politics of creditworthiness’, demands actuarial fairness in credit allocation and focuses critical attention on norms of objectivity, accuracy and transparency in predictive practices. It calls for an unbiased implementation of an actuarial model of fairness in credit markets and demands that everybody be treated in accordance with their ‘objective’ creditworthiness and risk, irrespective of ‘race’. This politics of creditworthiness has achieved widespread legibility and legitimacy and draws on a long political tradition of contestation. While I focus here primarily on the politics of risk and prediction in the run-up to the subprime crisis, my argument therefore extends beyond this example to a broader critique of the way in which actuarial conceptions of fairness have shaped the politicization of predictive practices.

While the politics of creditworthiness offer significant political resources, such as a critique of privatized and arbitrary forms of predictive power, my aim here is not to advance a straightforward vindication of the politics of prediction that I reconstruct. Instead, I seek to critically evaluate this dominant mode of contesting predictive power. My aim is to highlight some dangerously shallow waters into which it tends to run, and to make visible how it shapes dominant but narrow conceptions of racial (in)justice and contributes to the displacement of demands for redress.

I argue that the dominance of a politics of creditworthiness demonstrates the continued and underexamined influence of a liberal-proprietary conception of risk. By focusing on the unbiased implementation of actuarial fairness—i.e. the principle that everybody ought to be treated according to ‘their’ risk—an anti-racist politics of creditworthiness unwittingly affirms an individualist-proprietary conception of risk. The origins of this individualist-proprietary view of risk, I argue, can be traced to the emergence of modern predictive practices and, specifically, to the emergence of statistical individuality as a central principle of allocation and governance in the nineteenth century.Footnote 7

Crucially, this implicit affirmation of individual responsibility displaces more radical critiques that seek to foreground an institutional-historical analysis of the political economy of racialized market structures. By attributing moral responsibility to the individual for the risk they ‘represent’ by virtue of their statistical individuality, the social and historical processes that have distributed this risk unequally disappear from view. This obscures how race—as a political project, jointly pursued by the racial state and the ‘private’ real estate sector in the case of the housing-finance market (Glotzer, 2020; Gotham, 2014; Helper, 1969; Jackson, 1985; Taylor, 2019)—has systematically shaped the economic risks to which individuals are differentially exposed.Footnote 8 A racialized distribution of economic risks consequently appears as ‘fact’ without author or history. In failing to unsettle individualist-proprietary conceptions of risk, the politics of creditworthiness fails to effectively challenge the pervasive disavowal of past injustice. Paradoxically, then, demanding that individuals be treated in accordance with ‘their’ creditworthiness undermines the recognition of a debt that has gone unrecorded for too long: the debt of reparation and redress.

My analysis of the dynamics at play here is informed by—and makes a contribution to—recent debates about ‘racial capitalism’ (Gilmore, 2017; Jenkins & Leroy, 2021; Melamed, 2015; Robinson, 2000 [1983]; Virdee, 2019). By ‘racial capitalism’ I intend, following Jodi Melamed’s definition, ‘a thread of emergent critical understanding, proceeding from the recognition that procedures of racialization and capitalism are ultimately never separable from one another’ (Melamed, 2015, p. 77). On the one hand, my analysis makes visible a more radical critique of the racialized political economy of the housing-finance market that understands market-making and race-making processes as co-evolving and mutually articulated. It demonstrates the existence of a political tradition of contestation that, without explicitly deploying the language of ‘racial capitalism’, shows us how to conceptualize and articulate the racialization of finance beyond a framework of discrimination. At a deeper level, however, my analysis also demonstrates that these historical-institutional critiques of racialized market structures are often deeply entangled with, and insufficiently differentiated from, dominant anti-racist discourses that foreground a discrimination framework. This anti-discrimination frameworks conforms to a liberal understanding of the relationship between the structures of capitalist markets and race making, and undermines more radical strands of critique. This reveals how ‘racial capitalism’—as an ‘activist hermeneutic’ (Melamed, 2015, p. 76) that seeks to move us from a discrimination to a racialized market structures framework—is displaced by a paradigm that reinforces conceptions of racism as ‘residual’.Footnote 9 My analysis also shows how conceptions of ‘economic rationality’, abstracted from its specific historical-institutional form, and equally abstract conceptions of market discipline, are used in order to disavow responsibility for an unjust past. It therefore helps to expand the remit of debates about racial capitalism beyond the question of how market structures have been racialized to analyses of how markets foreclose redress.

The article will be structured as follows: the first part sketches how a proprietary relationship to predicted outcomes developed with the emergence of statistical individuality in the nineteenth century. I argue that a conception of actuarial fairness-as-justice emerged out of an actuarial rearticulation of the liberal-individualist paradigm of risk, which took place under the system-specific pressures of for-profit insurance practices. This rearticulation gave rise to successive political struggles that contested the articulation of categories of social difference in and through predictive practices. I argue that two dominant modes of anti-racist critiques of actuarial predictive practices emerged in struggles against the use of race as a proxy for financial risk, namely a rights-based approach that conceptualizes actuarial classifications as a violation of individual rights, and a ‘politics of classification’ (Krippner & Hirschmann, 2022) that politicizes predictive practices by challenging their objectivity and accuracy. The ‘politics of creditworthiness’ inherits its conceptual and normative framework from the latter.

The second part turns to the specific case study of debates about racial discrepancies in subprime lending in the run-up to the financial crisis in order to critically examine both the ascendancy of a politics of creditworthiness and to demonstrate its limits. In the section ‘Democratizing Credit? Costless Justice and Belated Inclusion’, I argue that advocates of financial deregulation portrayed the emergence of subprime lending as a ‘democratization of financial markets’. This, I argue, amounted to an idea of ‘costless justice’—the notion that free, deregulated markets would provide remedies for extant racialized exclusions. That notion of costless justice, I argue, naturalized the racialized segmentations and exclusions which had been built into the American housing and housing-finance markets, thus transforming what ought to have been an object of redress into an a-historical idea of ‘belated inclusion’.

The vision of subprime lending as a democratization of markets was contested from the beginning. In the next section ‘Seeing Like an Activist’,Footnote 10 I show how anti-predatory lending activists challenged this rose-tinted and a-historical view of the housing-finance market and analyse how they sought to problematize and politicize financial predictions of creditworthiness. I argue that they combined two strands of critique. On the one hand, they sought to draw attention to the ways in which red-lining and its aftermath had produced racialized market segmentations that resulted in a racialized landscape of vulnerability to exploitative lending practices. The most developed versions of this historical-institutional critique of racial inequality in the housing-finance market argued that deregulation, in combination with a market structure that was deeply shaped by red-lining and its afterlives, meant that structures of profitability had been de facto racialized. On the other hand, however, the dynamics of the debate about racial discrepancies in subprime lending increasingly undercut and marginalized the radical potential of this critique. As the debate came to focus increasingly on whether racial discrepancies in subprime lending reflected or exceeded pre-existing financial risk characteristics, such as income and credit history, the critique that became most legible demanded that everybody be treated in accordance with ‘their’ risk. In other words, a politics of creditworthiness, which conformed to a liberal politics of risk and attributed responsibility for ‘one’s’ risk to the individual, became dominant. I explore how this modality of contestation, while politically powerful, also limited what could be envisaged and demanded. In particular, I am concerned with the way in which the ‘formatting’ of the complaint in a language that was legible within the strictures of an actuarial fairness paradigm produced an a-historical understanding of the economic realities of the processes of racialization, and construed racialization and economic rationality as inherently contradictory logics and processes. I argue that this politics of creditworthiness ultimately failed to effectively challenge the paradigm of ‘costless justice’ and ‘belated inclusion’ that continued—and continues—to dominate debates about past and present racial injustice.

Contesting predictive practices

Although contemporary debates sometimes seem to suggest otherwise (cf. Weinberg, 2022, p. 76), both the general political problem of how to regulate the power to predict and classify, and the specific question of how to govern the way in which predictive practices employ and articulate categories of hierarchically ordered, ascriptive social difference, are constant rather than novel features of modern predictive practices (cf. Hacking, 2016; Harcourt, 2007; Levy, 2012; Simon, 1988). They emerge with the development of statistical individuality as a central mode of governance and a key principle of allocation in the nineteenth century. Modern predictive practices establish individual risk expectations based on statistical group averages, and pioneer the attribution of individual responsibility for statistically determined risk expectations. Beginning in the 1820s, advances in collecting statistical data at a scale previously unseen, and the increasingly widespread adoption of probabilistic methods of risk assessment, allowed commercial, for-profit institutions, such as insurance companies, to issue economically viable individualized predictions and risk assessments.Footnote 11 This enabled a differentiation between ‘populations’ with distinct ‘risk profiles’, and inaugurated a new way of linking individuality and group belonging.

Crucially, it introduced a basic, defining tension between how risks were predicted (at the aggregate level) and how (financial) responsibility for risks was being attributed (on an individual level), and it imbued predictive practices, especially the determination of risk categories, with political significance. Holding individuals responsible for the risk they represent by virtue of being part of a risk category thus gave rise to a distinctive political problem space in which the fairness of being held responsible for one’s statistically predicted outcomes was—and is—an object of political contestations.

Since the nineteenth century, two political answers have been developed in response to this basic tension: first, a social-democratic one, which essentially transcends this tension by decoupling individual contributions from individual risk expectations or probabilities (Ewald, 2020 [1986]); second, a liberal risk regime that retains this basic tension, and essentially treats individual risk expectations as a form of private property, thus establishing an actuarial rearticulation of the liberal notion of individual responsibility for risk.

Much of the existing social-theoretic and historical literature on the history and politics of risk has focused on the former, and has told the history of ‘risk paradigms’ as a shift from liberal-individualist approaches to risk to solidaristic risk-sharing paradigms, a shift that is usually held to have taken place during the late nineteenth and early twentieth century (Baker & Simon, 2002, p. 2; Ewald, 2002, 2020, pp. 273, 277ff; see also Hacker, 2006, p. 5; Moses, 2018; Witt, 2022). In the U.S., however, this shift is more aptly conceptualized as the co-existence of distinct risk paradigms within the same political order, often structuring different fields of social and economic action and representing precarious settlements rather than a decisive shift between solidaristic and liberal-individualist paradigms. For example, the New Deal introduced solidaristic risk-sharing practices in the fields of workers’ accident insurance and social insurance, while also maintaining predominantly liberal approaches to risk in health care and housing finance. In the U.S., therefore, the rise of actuarialism does not amount to a break with a liberal paradigm of individual responsibility, as is commonly asserted in Foucauldian risk studies (cf. Behrent, 2010; O’Malley, 1996) but is better understood as an actuarial rearticulation of a liberal paradigm of individual responsibility for one’s risk.

This rearticulation has given rise to persistent challenges to actuarial practices and the paradigm of actuarial fairness-as-justice. One prominent critique of the principle that individuals ought to be responsible for the risk they represent has taken the form of a rights-based legal challenge to actuarial logics. In this form, it has been articulated as a clash between individual rights and statistical-actuarial classifications of groups (Simon, 1988, pp. 776–782). On this understanding, the basic problem is that the individual is treated as a member of a group rather than on their own merits. This approach remains one of the prevailing ways of understanding what is wrong with actuarial fairness in legal and political thought and practice, including in contemporary debates about algorithmic fairness (cf. Hoffman, 2019).

Another dominant way of contesting actuarial practices—one that has been equally influential in the anti-racist contestation of economic calculative and predictive practices during red-lining and its afterlives—is a ‘politics of classification’ (Krippner & Hirschman, 2022) that does not challenge the basic principle of actuarial fairness but instead contests the categories and the epistemic procedures of actuarial predictive practices. This mode of contestation does not challenge actuarial predictive practices as a violation of individual rights but instead contests their future-making capacities on their own terrain by disputing the accuracy and objectivity of actuarial predictive practices. This, I argue, is the political tradition in which we should understand the politics of creditworthiness.

This modality of critique and contestation first became influential in anti-racist politicizations of prediction in the late nineteenth century, during the first boom of life insurance in the U.S., which also pioneered new forms of employing and articulating ideas of racial difference and racial hierarchy in predictive practices (Bouk, 2015).Footnote 12 In a wave of successive challenges to the life insurance industry, early civil rights activists, many of whom were associated with the short-lived Afro-American League, contested the explicit use of racial classifications in the calculation of expected life spans, and, consequently, insurance premiums (Alexander, 2011, pp. 44–45). Contesting economic predictive practices by engaging in a politics of knowledge production was also influential in successive struggles against red-lining and its afterlives that challenged the use of ascriptive racial classifications as predictors of risk and future economic value in the housing-finance market, such as the NAACP’s early campaigns against red-lining, as well as the Community Reinvestment Movement’s struggles against the aftermath of red-lining (Marchiel, 2021). In the second part of this article, I will show how this particular mode of contestation became dominant in debates about racial discrepancies in subprime lending, and explore the promises and pitfalls associated with the politics of creditworthiness.

Democratizing credit? ‘Costless justice’ and ‘belated inclusion’

When subprime lending first emerged as a result of the deregulation of financial markets and the rise of securitization in the 1990s (Engel & McCoy, 2010, p. 16), it was feted by many—including the then-Chairman of the Federal Reserve, Alan Greenspan, and Eugene Ludwig, then Comptroller of the Currency—as the ‘democratisation of credit’ (Greenspan, 1997; Ludwig, 1996). The sophistication of a deregulated financial industry, with its heightened agility and ability to hedge against risks and uncertainties, it was claimed, would benefit many who had traditionally remained locked out of the credit economy—especially ‘minorities’, and low-income borrowers (Greenspan, 2007, pp. 229–230). Hedging against risks in ever more complicated and ingenious ways would allow lenders to offset the risks that these ‘new’ financial subjects supposedly represented. In a hearing before the House Committee on Banking and Financial Services, Ralph Rohner, the Special Counsel to the Consumer Bankers Association, asserted, for example, that ‘[t]he data on mortgage volumes, including the encouraging figures on loans to minorities and in previously under-served communities, are a source of great pride to the banking industry’. He approvingly noted that it ‘has been called by some the democratization of credit’ (Predatory Lending Practices, 2000, p. 563). Similarly, then-chairman of the Mortgage Bankers Association, Rob Couch, testified before the Committee on Financial Services, ‘[i]n recent years the mortgage banking industry has greatly expanded its efforts to reach families who traditionally lacked access to credit. Many innovative credit options have made it possible for millions of low- and moderate-income families to build their family’s wealth through home ownership. In 2001, for example, minorities accounted for about 32% of first-time homebuyers, up from only 19% as recently as 1993. The Federal Reserve Board’s Governor Edward M. Gramlich calls this a true democratization of credit’ (Protecting Homeowners, 2003, p. 24).

The deregulation of financial markets was, as Greta Krippner has convincingly argued, the consequence of a political response to the distributional conflicts that emerged in the 1960s; a response that explicitly sought to abdicate responsibility for politically costly decisions about resource allocation to markets (Krippner, 2011, p. 58ff). Narrating the expansion of access to credit, which came in the form of more volatile and expensive forms of credit, as a ‘democratization’ was therefore particularly ironic: it was, in essence, the tail end of a de-democratizing devolution of decision-making power to the market.

But what is also notable in this narrative of deregulation as democratization, and subprime lending as a means of belated inclusion, is that racialized exclusions and asymmetries within the mortgage market were simply taken for granted. As the comments above symptomatically signal, there was little reflection on the historical roots of racialized exclusions in the mortgage market. Note, for example, the collapse of the distinction between ‘low-income’ and ‘minorities’ in Rob Couch’s remarks. Similarly, the widespread use of the term ‘minorities’ as the marker of a disadvantaged subject-position not only failed to name a relationship of domination rather than of numerical difference (Wingrove-Haugland & McLeod, 2021, p. 1). It also posited a generalized insider–outsider dynamic as the source of disadvantage (cf. Valls, 2018, p. 5), which obfuscated the specific histories of processes of racialization that had shaped the mortgage market.

In my view, this conception of subprime lending as a means of democratization and belated inclusion is best characterized as a narrative of ‘costless justice’ that represents market flexibilization as redress. In its breathless excitement for politically cheap market-based ‘solutions’ (Krippner, 2011, p. 59) to persistent racialized inequalities, the celebration of market flexibilization as belated inclusion, and belated inclusion as redress, turned the afterlives of red-lining into unspecified exclusions, without history or author, and in need of neither explanation nor repair.

At the same time, newly flexibilized technologies of debt essentially facilitated a ‘responsabilization’ (Joseph, 2014, p. xi) that allowed financial institutions to ascribe responsibility for the ‘risk’ these ‘new’ financial subjects supposedly represented. This was the latest iteration in a long line of for-profit socio-technical risk commodification techniques that allowed for the transformation of structural-historical location into personal responsibility. Crucially, the responsabilization of subjects who were supposedly ‘included belatedly’, also involved a de-responsibilization, a divestment from responsibility for redress for the beneficiaries of past injustice. This effectively transformed an outstanding debt of redress into a debt to be paid by those subject to the legacies of sedimented racism in the housing-finance market. The for-profit assessments of risk and creditworthiness, and the legal and regulatory protocols governing the appropriability of the upsides of commodified risk, generated formal debt claims that produced a sense of individual responsibility and simultaneously obfuscated the historical origins of a racialized distribution of economic risks.

This double movement of responsibilization and de-responsibilization, which was accomplished through the production of formal claims of debt and credit, was entangled with what Sadiya Hartman, in a different historical context, has called the ‘figurative accrual of debt’ (Hartman, 1997, p. 132). The ‘figurative accrual of debt’ refers to a set of polyvalent, and sometimes contradictory, cultural narratives about who owes what to whom, in which the financial and the normative are closely interwoven. Abstract actuarial assessments of creditworthiness engendered a contradictory ‘amnesic vision’ (Hartman, 1997, p. 133) of history that neutralized and naturalized histories in need of redress. It simultaneously denied that injustice had taken place and positioned those disadvantaged by past injustice as responsible for rectification and redress. Insofar as for-profit creditworthiness assessments contributed to the transformation of the risks to which one is exposed by virtue of one’s structural-historical position into risk as personal responsibility for which one can be held financially responsible, they clouded the histories of the imbrication of racialization and capitalist market institutions, and thus, demands for redress. In other words, for-profit ‘calculative devices’ played a crucial role in ‘forgetting’ the history behind uneven access to credit, and, more fundamentally, the history and politics of a racialized exposure to economic risks. This was an ‘active forgetting’ along the lines of what Charles Mills has termed ‘white ignorance’ (Mills, 2007): a forgetting that depends on disavowing what is known in order to construct and maintain a notion of ‘white innocence’ (Wekker, 2016).

It is this obfuscation of a history in need of redress that allowed for the narration of market flexibilization as ‘belated inclusion’, and ‘belated inclusion’ as sufficient redress or even as a ‘gift’ (Hartman, 1997, pp. 130–136). In other words, it allowed for the transformation of a debt owed to those subject to the legacies of racial exclusion and domination into a debt owed by those subject to the legacies of racial exclusion and domination.Footnote 13

Seeing like an activist

See endnote 10.

From the start, fair lending organizations systematically challenged this idea of costless justice with its rose-tinted vision of the market’s capacities for integration, and its implicit equation of inclusion (on unequal terms) and redress. It was true, activists acknowledged, that the rise of subprime lending in the late 1990s had radically changed the landscape of credit. Black and brown neighbourhoods that mortgage lenders had previously shunned as ‘too risky’ suddenly became inundated with subprime mortgage capital (Dymski, 2009; Rugh & Massey, 2010). But activists challenged the portrayal of this development as a form of redress for existing exclusions. Instead of rectifying the racialized segmentations of the mortgage market by integrating those previously excluded, they argued, subprime lending reproduced, intensified and exploited the market segmentations that had been established by red-lining and its afterlives. Subprime lending, according to their critique, constituted a reverse replication of the histories of red-lining rather than the antidote to the exclusions and segmentations in financial markets.Footnote 15 If subprime lending did effectuate a form of integration, activists argued, it was an integration into the market on unequal terms, an intensification of predatory and exploitative lending that made use of existing vulnerabilities that were the result of histories of injustice.

In the late 1990s and early 2000s, fair lending associations, such as the Association of Community Organizations for Reform Now (ACORN), and the National Training and Information Center, published reports detailing the calamitous results of subprime lending for communities, and the disproportionately negative impact that subprime lending was having in predominantly Black and Latinx neighbourhoods (ACORN, 2001). An emergent anti-predatory lending movement began to systematically push for a national conversation about subprime lending and its entanglements with the afterlives of red-lining. In 1999, for example, the National Training and Information Center (NTIC), a Chicago-based fair lending organization, published a report about the devastating impact of predatory subprime lending in the Chicago area (Nagazumi et al., 1999, p. 5). The report highlighted the rapid increase of subprime lending and a resulting jump in foreclosures and the abandonment of homes. The NTIC report did not explicitly analyse the racial dimensions of predatory subprime lending but it inspired studies that did. In November, the Woodstock Institute published ‘Two Steps Back: The Dual Mortgage Market, Predatory Lending, and the Undoing of Community Development’, which documented the racial discrepancies in subprime lending in Chicago. It found that 52.9% of refinance loans made in predominantly African-American neighbourhoods were subprime loans, as compared to 9.39% of refinance loans in predominantly white neighbourhoods (Immergluck, 1999). A report published by the Department of Housing and Urban Development (HUD) in November of 1999 found similar discrepancies at the national level (Scheessele, 1999).

By the early 2000s, the problem of predatory and exploitative lending practices in the subprime industry, and racial discrepancies in the incidence of subprime lending, had become sufficiently pressing to occasion numerous Congressional hearings, as well as a joint HUD-Treasury anti-predatory lending task force. In March 2000, for example, Sen. Chuck Schumer commissioned a report on racial lending patterns in New York City for a Congressional Hearing. The study concluded that ‘as far as lending practices in New York City are concerned, blacks and whites may as well be living on two different planets’ (Predatory Lending Practices, 2000, pp. 9, 128). The HUD-Treasury anti-predatory lending task force came to a similar conclusion, finding disproportionally high concentration of subprime lending in majority black neighbourhoods in Chicago, Atlanta, Los Angeles, Baltimore, and New York City (HUD, 2000).

In contrast to ideas of ‘costless justice’, fair lending activists put forward a historical analysis of the subprime lending market that highlighted the afterlives of red-lining. They argued that subprime lenders, in an inversion of red-lining, were actively targeting black and brown neighbourhoods for exploitative and predatory lending practices. As Gloria Waldron, a member of the Association of Community Organizations for Reform Now (ACORN) put it in a hearing before the Banking and Financial Services Committee, ‘unfortunately, the statistics confirm what is obvious to us. These lenders target our communities’ (Predatory Lending, 2000, p. 63).

At their most incisive, these critiques of the operations of financial markets did not allege that exploitative lending practices were solely or even primarily distorted by extra-economic racial prejudice. Instead, fair lending activists put forward a critique in which historically emergent structures of profitability incentivized racialized exploitation. This critique saw the housing and credit market as deeply shaped by what one might term a public–private racial project in ways that allowed for the exploitation of racialized market segmentations. It explained the exploitation of racialized market structures not as a deviation from economic rationality but as an instantiation of racialized economic incentives. This critique made two aspects of the problem visible: First, it highlighted that racialized market segmentations were traceable to a racial project of segregation and subordination (of which red-lining was but one aspect) and to the persistent failure to offer meaningful redress (cf. Taylor, 2019). Second, fair lending activists correctly identified that securitization had changed the risk orientation of subprime lenders, and that this made it profitable for subprime lenders to exploit racialized asymmetries of access to credit. Securitization meant that mortgage originators no longer retained issued loans on their portfolios. Instead, loans were sold on to investment banks or other arrangers, where they were aggregated and resold as tranches. This shifted the emphasis from a careful selection of risks to a maximization of the volume of mortgage originations (cf. Engel & McCoy, 2010, pp. 28, 40-41). Given pre-existing racial market segmentations attributable to the history of red-lining and its afterlives, including the abandonment of so-called ‘inner city’ neighbourhoods by mainstream banking services in the 1970s and 1980s, this change in the risk orientation of lenders, combined with weakened regulatory oversight, allowed for the systematic exploitation of the racial segmentation of the mortgage market (cf. Immergluck, 2015). Fair lending advocates and their political allies argued that lenders recognized this market segmentation, and strategically exploited the resulting vulnerabilities. In her opening statement in the 2003 hearing on ‘Protecting Homeowners: Preventing Abusive Lending While Preserving Access to Credit’, Maxine Waters directly accused lenders of exploitative practices: ‘These lenders are able to engage in predatory activities because credit-starved communities – unfortunately, usually minorities and elderly persons – have little access to traditional sources of credit’ (Protecting Homeowners, 2003, p. 16). In a hearing on Predatory Lending Practices in May 2000, William Brennan from the Atlanta Legal Aid Society put it more forcefully, arguing that ‘the financial service industry has created a system of financial apartheid and have done it on purpose. … People with A credit by and large are treated very well by our financial institutions and banks. People with C and D credit are purposely abused for the sake of exploitation and profits … They target minority groups because they have historically been cut out of access to credit and these lenders know that’ (Predatory Lending Practices, 2000, p. 58).

Anti-racist, fair lending activists therefore painted a picture of the subprime lending market that was a far cry from the idea that the deregulation of financial markets constituted a ‘democratization of credit’. They drew attention to the ways in which the housing market had been deeply segmented in the afterlives of red-lining in ways that produced an asymmetrical power landscape that allowed for the deployment of particularly exploitative practices in black and brown neighbourhoods. Where supporters of the subprime lending industry saw efficient calculative practices that correctly assessed uneven risk distributions, and more flexible loan arrangements that offered ‘riskier’ financial subjects a step unto the ladder of American homeownership, fair lending activists argued that the calculative and predictive practices that ostensibly served to assess financial risk had been systematically compromised by the racialized structure of the housing-finance market. They argued that predictive practices that were supposed to assess financial risks functioned more like justificatory ornamentation, fuelling an economy of ultra-processed debt in which the upsides of risks could be profitably appropriated while the downsides could be passed on to investors on the secondary market and to the borrowers themselves.

Unlike many of the critiques that emerged in the wake of the subprime crisis, this critique was not simply directed against the hubris or inaccuracy of the predictive practices at play (cf. Derman, 2011; Rebonato, 2007). Instead, its analysis of the crisis understood the racial segmentation of the housing-finance market and ongoing practices of racialization as crucial aspects of the crisis, and articulated a powerful critique of racialization as a contributory force in the historical development of the structures of profitability.

From the afterlives of red-lining to a politics of creditworthiness

However, over the course of this debate, the radical potential of this critique was undercut by the dominant politics of creditworthiness. Historical-institutional critiques of the racialization of the structures of the housing-finance market existed in an unacknowledged tension with demands for ‘accuracy’ and ‘objectivity’ as the norms governing fair predictive practices. As advocates of deregulation and professional associations representing the banking sector sought to rebut charges of racialized predatory lending practices, political debates about racial discrepancies in subprime lending increasingly came to focus on the question of whether subprime lending offered a rational and efficient market response to a segment of risky borrowers or whether racial discrepancies in subprime lending exceeded pre-existing financial risk characteristics, such as income and credit history (cf. Predatory Lending Practices, 2000; Protecting Homeowners, 2003; Subprime Lending, 2004). Debates revolved around the question whether risk was being measured and attributed fairly according to established standards of risk measurement. Were black and brown borrowers treated differently than identically situated white borrowers? Could the racialized distribution of subprime lending plausibly be interpreted as an adequate reflection of a pre-existing distribution of financial risk factors? Or did subprime lenders, in an inversion of red-lining, steer black borrowers into subprime loans? A virtual flood of NGO and government reports, as well as countless academic articles, rushed to address this question.

In trying to ascertain the correspondence of subprime lending with a ‘pre-existing’ distribution of financial risk, the disagreements over the nature of the subprime market crystallized around three different aspects: First, whether it was possible to know if racial discrepancies in subprime lending corresponded to a pre-established distribution of underwriting criteria; second, whether market actors were incentivized, willing and competent enough to ‘get risk right’; and finally, who, if anybody, had the legitimacy and capability to sanction or challenge the risk assessments made by market actors. Today, most scholars agree that the racialized distribution of subprime lending cannot simply be explained in terms of a pre-existing risk characteristic distribution, such as income or credit history. But, ironically, the framing of the debate around this question turned out to be more consequential than the answer to the question. Debates about how to ‘get risk right’ had a little noted side effect: they produced an implicit commitment to actuarial conceptions of fairness-as-justice and a proprietary-individualist conception of risk. In other words, the framing of the debate proved productive: It increasingly made the objective assessments of creditworthiness the central object of contestation, and established objective assessments of creditworthiness and accurate predictions of default risk as the basis on which to evaluate the commodification of risks and futures.

In seeking to trouble the notion that the existing subprime lending market was efficient and assessed financial risks accurately and objectively, many fair housing advocates evoked treatment according to one’s creditworthiness as the standard that needed to be met in order to ensure fairness (cf. Morris (NAACP), 2010, p. 3; NCRC, 2003). Notwithstanding substantive disagreements, the structure of the debate elevated the existing distribution of financial risk characteristics to a standard of fairness. The debate was not about whether being treated in accordance with one’s risk characteristics constituted an adequate normative principle, but rather what regulatory interventions were necessary in order to ensure that borrowers were treated in accordance with their creditworthiness. This implicitly endorsed an actuarial conception of fairness-as-justice in risk commodification as the basis for evaluating claims of racial injustice. As the title of one congressional committee hearing on subprime lending put it, the imperative was to get borrowers ‘the credit they deserved’ (Helping Consumers, 2005, p. i). The ‘credit they deserved’, however, was the credit borrowers ‘deserved’ according to their underwriting characteristics.

Despite some of its strengths, the politics of creditworthiness was caught in a recursive loop that transformed a demand for redress into a demand for inclusion on ‘objective’ terms, and turned a critique of the gravitational pull of racialized market structures on predictive practices into a critique of bias in predictive practices. Insofar as it challenged lending institutions’ predictive practices, it advanced a critique of private and arbitrary predictive power and raised the question of who gets to judge what future is worthy of being made (i.e. who is judged to be creditworthy). But the structure of the debate narrowed this question in a politically invidious way. More radical critiques that had essentially identified the ways in which the very structures of the housing market—due to their long-term entanglements with political projects of race making—had been shaped so as to ‘corrupt’ economic rationality were displaced. Anti-racist modes of politicizing prediction were effectively caught between revealing the ways in which racial practices and logics, both past and present, continued to shape the housing-finance market, and recourse to a moral economy of risk that obfuscated those very structures by appealing to formally equal treatment in accordance with individual creditworthiness, thus rendering the ongoing legacy of past racial injustice invisible. Moreover, under the cover of a struggle over knowledge production, the debate unwittingly resurrected a ‘residual’ understanding of racism that construed the relationship between race and economic rationality as opposed and antagonistic. This focused public attention, and moral condemnation, on so-called ‘independent race effects’, i.e. those aspects of racial discrepancies in subprime lending that could not be attributed to any pre-existing discrepancies in underwriting characteristics. It, too, displaced debates about how such pre-existing discrepancies in underwriting characteristics had come to exist in the first place, and managed to drown out more radical conceptions of ‘race’ as a form of political and social power that structures property relations.

This displacement, however, leaves one without the normative or conceptual resources to tackle the ways in which past and present racial injustice has shaped—and continues to shape—the existing distribution of financial risk and creditworthiness. It narrows the scope of critique, and fails to see race as a mode of power that mediates economic processes. Sustaining a more radical critique would require a more thorough break with the alchemy that turns injury and injustice into individual responsibility. It would entail foregrounding the normative question that the politics of creditworthiness tends to displace, namely: If the pre-existing distribution of financial risk and creditworthiness is itself racialized due to past racial injustice and the racial segmentation of markets, why should those that have been disadvantaged most by these societal processes bear the cost of this history? And what are meaningful remedies for the ways in which racial injustice has shaped the present distribution of financial risk and creditworthiness?

The limitations of a politics of creditworthiness show that the politicization of prediction ought to move from a liberal regime of risk to more solidaristic frameworks for risk sharing. In other words, this critique could open the door to more expansive ways of politicizing risk and creditworthiness that do not simply seek to predict default risk and future profitability more accurately but instead demand solidaristic and reparative forms of risk sharing. At the very least, this would include challenging the state’s reliance on credit as a major form of social provisioning. Politicizing prediction, then, should expand beyond the question of what features of the past should and do matter for the projection of risk and future profitability, about who is competent and incentivized to realize the implementation of accurately predictive apparatuses, and centre the question of what risks ought to be socialized.

Conclusion

Predictive practices are modes of future-making, and we must find effective ways of politicizing the futures they are making. Like some of the critics of the limitations of recent debates about algorithmic fairness (Davis et al., 2021; Hoffmann, 2019), I have argued that we must go beyond the goal of accuracy, objectivity, and fair representation in prediction in order to effectively disrupt how predictive practices are re-making race and racialized economic inequality. However, to mobilize effectively against a narrowing of the ambit of anti-racist politics, we need to be aware of the deep roots that actuarial fairness has in hegemonic political imaginaries, including in oppositional discourses, and attend to the dynamics of debates that consistently reproduce an attachment to actuarial fairness and promote proprietary-individualist conceptions of risk.

I have here argued that an attachment to actuarial fairness and individualist-proprietary conceptions of risk continues to reverberate in anti-racist politicizations of predictive practices and undermines efforts to make visible the racialization of market structures, foregrounding, instead, a framework of anti-discrimination. This fails to effectively challenge an ‘amnesic vision’ (Hartman, 2019, p. 133) of history which obscures past injustice and renders demands for redress illegible.

The immediate political problem this raises is how to disrupt the production of this particular mode of moralizing risk that inverts the historical relationship of responsibility for redress. My analysis suggests that anti-racist politics of prediction, rather than contesting predictive practices on the basis of actuarial fairness, should break decisively with proprietary-individualist conceptions of risk and seek to refuse the moralizing effects of responsibilization—i.e. seek to refuse the norm of ‘to each according to their risk’. In the decades since the financial crisis of 2007/2008, movements and organizations such as Strike Debt and the Debt Collective have advanced a political project that explicitly refuses the moralizing effects of formal debt claims; this could potentially serve as a model for how to refuse the responsabilization for ‘one’s’ risk. Any attempt to do so, however, will face strong political headwinds, not least due to the gravitational pull that private, for-profit predictive practices continue to exert on politicizations of prediction, and a legal framework in which profitability too often constitutes the outside of possibilities for redress. Zooming out, these difficulties of articulating a politics of risk, debt, and prediction outside of a hegemonic liberal-individualist political rationality illustrate and invite broader questions about the capacity of the liberal tradition writ large to effectively address legacies of enduring racial injustice.Footnote 16

But my article also has implications beyond the politics of prediction. In highlighting the role that market-based calculative and predictive practices—and the diffusion and inversion of political responsibility that they enable—play in foreclosing redress and sustaining a project of ‘white innocence’ (Wekker, 2016), it can enrich recent debates about racial capitalism. The question that my analysis here opens up is how for-profit predictive practices have shaped our political imaginary, including our conception of anti-racism. I want to conclude by suggesting that this is part of a longer and broader history of how a conception of capitalism and racism as oppositional rather than imbricated has become hegemonic. This also opens up a broader question for debates about racial capitalism: namely, how market institutions contribute to the obfuscation of the history of their own racialization. This, in turn, would expand the remit of debates about racial capitalism from the historical racialization of market institutions to the foreclosure of redress for past injustice through private control over capital and the diffusion of responsibility that is inherent in capitalist markets.