ArticlePDF Available

‘All data is credit data’: Constituting the unbanked


Abstract and Figures

Global financial and data capitalism has constituted new forms of knowledge, novel inscriptions which make that knowledge tangible and new ways of visualizing sources of value and profit. This paper examines a cluster of new practices designed to make visible – and extract value from – those without formal credit scores in contemporary financial markets. Many ‘financial inclusion’ projects now attempt to score the ‘credit invisible’ by drawing on a range of alternative data – non-financial payment streams, academic records, behavioural signals gleaned from online or social media footprints and results generated via digitized psychometric testing – and by assessing that data in relation to models of risk assessment based on the analysis of big data. I argue in this paper that these experiments in alternative credit scoring constitute the unbanked as an important, and dubious, category of knowledge and intervention. I also argue that attempts to score the unbanked offer a revealing glimpse of many of the social and political limitations associated with projects of ‘inclusion’. Although often imagined as forms of pristine incorporation, inclusion projects often constitute troubling new kinds of social sorting and segmentation.
Content may be subject to copyright.
‘All data is credit data’:
Constituting the unbanked
Rob Aitken
University of Alberta, Canada
Global financial and data capitalism has constituted new forms of knowledge, novel inscriptions which
make that knowledge tangible and new ways of visualizing sources of value and profit. This paper
examines a cluster of new practices designed to make visible – and extract value from – those without
formal credit scores in contemporary financial markets. Many ‘financial inclusion’ projects now attempt
to score the ‘credit invisible’ by drawing on a range of alternative data – non-financial payment streams,
academic records, behavioural signals gleaned from online or social media footprints and results
generated via digitized psychometric testing – and by assessing that data in relation to models of
risk assessment based on the analysis of big data. I argue in this paper that these experiments in
alternative credit scoring constitute the unbanked as an important, and dubious, category of knowledge
and intervention. I also argue that attempts to score the unbanked offer a revealing glimpse of many of
the social and political limitations associated with projects of ‘inclusion’. Although often imagined as
forms of pristine incorporation, inclusion projects often constitute troubling new kinds of social sorting
and segmentation.
Financialization, credit invisibles, big data, unbanked, inclusion/exclusion, visualization
Everyone must be counted, but only if they count. Dead migrants don’t count. The [migrant]
woman who drowned while giving birth was not a biometric subject, she was a biodegradable
one....But I want her to be known by more than just the number she was given after being
hauled out of the water—288 (and 289 for her baby)—because otherwise the story ...remains
infinitely reproducible to the point of abstraction. (Saunders, 2016: 10–11)
In December 2015 a series of large banks based in New York – JP Morgan Chase, Citibank,
Bank of America – confirmed their refusal to accept New York City’s new identification card
as a basis for new accounts. The ID programme was key to New York City mayor Bill de
Blasio’s efforts to address a growing population of ‘unbanked’ New Yorkers. Despite regu-
latory assurances, the large banks continue to assert their own right to determine what
documentation might best distinguish those who can access formal financial practices
Corresponding author:
Rob Aitken, Department of Political Science, University of Alberta, Edmonton, AB T6G 2R3, Canada.
Competition & Change
2017, Vol. 21(4) 274–300
!The Author(s) 2017
Reprints and permissions:
DOI: 10.1177/1024529417712830
from those deemed too risky. ‘These are business judgments’, notes Michael P. Smith, presi-
dent of the New York Banker’s Association, ‘it is ultimately the institution that has to pass
muster if there were ever a problem’ (quoted in Corkery and Silver-Greenberg, 2015).
The ongoing drama of de Blasio’s experiment foregrounds the tensions which attend the
‘unbanked’, a figure at the core of recent financial ‘inclusion’ talk. But where did this cat-
egory – the ‘unbanked’ – come from? What does it convene under its name? And what does it
signify for those who are assembled under its purview? Perhaps because it refers to a ‘lack’,
to the persistent absence of finance in particular lives, the category of ‘unbanked’ can often
seem like an unproblematic residue, a self-evident label for a real or obvious condition. The
contention at the heart of this paper, however, is that the unbanked is itself a category that is
produced and made real in particular kinds of ways.
Not just a self-evident category or a
label for some already existing problem or pathology, the unbanked is a method of social
sorting key to the ways in which the economic lives of precarious populations are ‘made up’
and rendered governable.
One site that is crucial to the genealogy of the unbanked is the practice of alternative
credit scoring. A series of experiments now proliferate in the use of alternative sources of
data in the formalization of credit scores for those without credit records or files. These
alternative data include local public records, social networking patterns, academic achieve-
ment records, mobile phone usage, non-financial payment histories, and, increasingly, psy-
chometric test results. These experiments, I argue, are an important setting key to the ways in
which the ‘unbanked’ is constituted as a category of knowledge and intervention.
I pay particular attention in this paper to the ways in which these attempts to score the
unbanked orbit around ‘regimes of visibility’, what Brighenti (2010: 3) calls the ‘complex
social, technical and political arrangements’ with which certain aspects of experience become
‘sensorially perceptible’. Like all bodies, before the unbanked can be governed, they must
first be made visible in particular ways, literally re-presented in forms which make them
amenable to intervention. Organized attempts to ‘score’ the unbanked are particularly pre-
occupied with a language of the visual. The unbanked, for example, are often most import-
antly framed as ‘credit invisible’, a population defined by their lack of legible trace within
any formalized mode of credit practice. Addressing the unbanked becomes, by extension, a
kind of exercise in making visible, of finding methods with which they are made perceptible;
an attempt ‘to recognize creditworthy individuals who would otherwise be difficult to iden-
tify’, and to record the ‘observable behaviour’ of those outside of formal credit records
(FICO, 2015: 1–9).
In this paper I argue that experiments in alternative credit data both constitute the
unbanked as a category of knowledge and offer a particular kind of financialization.
Although it has been associated with a diverse range of inflections, in this paper I draw
on a conception of financialization as a process of assemblage and an ‘act of configuration’ in
which forms of financial valuation are constituted in novel kinds of ways and out new kinds
of spaces and practices. By finding ways to render the credit invisible calculable, I argue,
alternative credit scoring experiments are attempts to constitute and extract financial value
from the places where it is invisible. This is a kind of financialization made possible by
finding new ways to know and see those without any legibility within mainstream credit
networks. This has required the construction of calculative infrastructures; the connective
tissues in a digital geography linking disparate sources of data to conventional credit scoring
bureaucracies. Although they are often framed as forms of financial ‘inclusion’, I argue
that alternative credit scoring experiments entail risks for those who are made newly visible.
Aitken 275
Even though alternative credit scores can provide access to financial services, for some, they
are also a kind of ‘social sorting’ designed to separate those that might profitably carry credit
from those who cannot.
To develop this kind of analysis, this paper is divided into three main sections. The first
section sets the theoretical context for the paper by offering a particular conception of
financialization not as reference to the intensifying power or finance capital per se but to
a process – to the ‘acts of configuration’ – that make certain objects, practices or bodies
identifiable and available as ‘financial’ assets or sources of financial value in the first place.
Drawing on Leyshon and Thrift and on Bruno Latour’s notion of inscriptions, this section
also places theoretical emphasis on the crucial ways in which objects are made visible and
visualizable. The second section enters into a more concrete discussion of these theoretical
insights by mapping the empirical contours of alternative credit scoring experiments. This
section emphasizes these experiments as attempts to constitute the unbanked as a calculable
body. This ambition requires the construction of calculative infrastructures that constitute
knowledge about the unbanked as ‘immutable mobiles’ and transport that knowledge to
locations where it can be used to generate financial value. A third section argues that this
knowledge is, like many forms of calculation, a kind of social sorting. Although it is framed
within a discourse of inclusion, alternative credit scoring often establishes forms of sorting
and segmentation. A conclusion returns to the complexities of financial ‘inclusion’ and
‘exclusion’ not as antinomies but as mutually implicated conditions.
Financialization, visibility, inscriptions
...the machinery for measuring, modeling, managing, predicting, commoditizing, and exploiting
risk has become the central diacritic of modern capitalism. Financial markets lead and shape
other markets, financial capital vastly outstrips manufacturing or industrial capital [and] finan-
cial policymakers dominate global economic policy. (Appadurai, 2016: 44)
Many of the recent conversations regarding the contemporary trajectory of capitalist devel-
opment have foreground the increasing complexity of financial markets. At the heart of this
complexity are the antinomies of what Martin et al. (2008: 121) refer to as decomposition
and recomposition; processes that both profitably package risk into bundled and ‘deloca-
lized’ instruments (securitization) and distribute risk widely among dispersed investors
These sophisticated practices – what Appadurai refers to as a machinery
designed to manage and commodify risk – have become a definitive feature of the global
political economy, the ‘leading edge of capital’ (Martin et al., 2008: 121) and have crystal-
lized a debate around the concept of financialization.
For Aalbers (2016: 2) financialization
refers to the ‘increasing dominance of financial actors, markets, practices, measurements and
narratives’ (see also Watkins, 2017).
At a general level this has often referred not only to the
increasing power of financial actors but also to the dramatic drift of financial techniques and
practices into new domains. Financial techniques of calculation are increasingly called upon
to help govern non-financial practices of all variety including education, social policy, health
care and the management of global climate change. ‘These calculative practices’, notes
Hiss (2013: 235) ‘enable the definition, determination, categorization, measurement and
standardization of objects (and particular processes) by linking them to the structures and
rules of financial markets’.
To provide some substantial depth to this general discussion Van der Zwan (2014: 100)
identifies three particular types of analyses of the ways in which ‘an increasingly autonomous
276 Competition & Change 21(4)
realm of global finance has altered ...the inner workings of democratic society’ (see also
Epstein, 2005: 3).
A first cluster of analysis frames financialization as ‘the tendency for
profit making in the economy to occur increasingly through financial channels rather than
through productive activities’ (Krippner, 2011: 4).
This novel form of accumulation is
marked by steady growth in the actual size and depth of the financial economy. In the
United States, for example, the financial sector has grown steadily between 1947 and
2006, a period which witnessed a dramatic growth in the proportion of GDP constituted
through financial activities (Philippon, 2008: 2; see also Aalbers, 2008: 149–151).
A second cluster of analyses of financialization pays particular attention to an ‘increased
shareholder value orientation’ (Van Treeck, 2009: 908). The maximization of shareholder
value and the consolidation of returns to shareholders have become paramount in the ways
in which firms are governed (Blackburn, 2006: 42–43).
‘The growing power of equity mar-
kets’, notes Montogmerie (2008: 237), ‘shifted management practices in publicly listed firms
from long-term gains through dividend payments toward short-term profit strategies based
on movements in share prices’. Moreover, many non-financial firms are now predominantly
oriented not to their own productive calculations but to the importance of financial criteria.
As they become oriented to shareholder value, non-financial firms hold an increasing share
of financial assets in their portfolios and increasingly seek out returns on financial invest-
ments as a key growth strategy. (Davis, 2016).
The emergence of shareholder value as a
basis of firm orientation has often entailed redistribution of power and material claims away
from workers and into the hands of or those with connections to financial portfolios.
A third broad conversation of financialization relates to how finance is experienced in the
practice, culture or knowledge of everyday life.
As Martin et al. (2008: 122) put it, finance
is ‘a machine for living whose architecture is at once international and intimate’ (see also
Langley, 2008).
At one level, focusing on the everyday intimacies of finance entails atten-
tion to the connections between working-class/everyday populations and financial markets
(Christopherson et al., 2013: 352; Harmes, 2001). At another level, however, situating
finance at the intersection of international pressures and the intimate vectors of everyday
life requires more finely grained attention to the complexities of everyday subjectivity and
identities, to the various processes which have given ‘rise of the citizen as investor’ (Van der
Zwan, 2014: 111). This process is evident, for example, in Langley’s now classical analysis of
the processes of identification associated with the consumer credit boom in the early 2000s.
In this boom, Langley (2008: 135) notes ‘prudence and thrift are displaced by new moral and
calculative self-disciplines ...the assembly of responsible and entrepreneurial borrowers’.
What attends financialization is a powerful set of cultural impulses which emphasize the
importance of a certain kind of ‘ownership’ society grounded not in the physical confines of
work or production, but in the crucible of financial exchange and entrepreneurial effort
(Christopherson et al., 2013: 354; see also Pathak, 2014: 91).
For Martin, finance is not
simply a mode of economic practice, but a ‘subjectivity and a moral code’ which ‘promises a
way to develop the self’ or a form of ‘self-mastery that channels doubt over uncertain
identity into fruitful activity’ (Martin, 2002: 8–9).
As Martin et al. (2008: 121) argue in ‘this financialized world, people are also invited to
treat their lives as if they were corporate spreadsheets parsed out and recombined bits of
their activity’. The relationship between finance and an investing self, however, is complex.
As Langley himself implies, the processes of subjectification are uncertain. To address this
ambiguity financialization research is imminently empirical and particular. ‘High level over-
views’, argue Montgomerie and Williams (2009: 104), ‘do not engage with the detail of the
Aitken 277
technical devices that channel funds or the social mechanisms of elite power and mass par-
ticipation’. Appleyard et al. (2016a, 2016b), for example, argue for a ‘variegated’ analysis of
the connection between credit networks and low income groups. In this analysis, ‘subprime’
borrowers are neither fully included not simply excluded from mainstream credit networks
but often move fluidly in ways that accentuate multiple sets of financial identities (Appleyard
et al., 2016a: 310).
Financialization research now explores a complex range of ways in which finance shapes
and connects the contours of everyday life and finance ‘writ large’ (Davis and Walsh, 2016:
666; Fernandez and Aalbers, 2016: 72; Nolke et al., 2013: 211)
; to help identify and assess
what Wark (2017) describes as a culture of ‘dependency on ...a spread of risks to be
optioned, swapped and hedged’. Theoretically, I want to build on this work by offering a
slightly more mundane conception of what it might mean to think of financialization as a
variegated practice: by assessing the quite concrete ways in which objects are rendered finan-
cial in the first place. Before financial markets can function, there must first be a prior
process which identifies and formats the kinds of assets that might serve as the basis of
financial circulation; ‘a source of value from which financial innovation can proceed’
(Leyshon and Thrift, 2007: 98). This requirement for sources of value entails the search
for new assets and the conversion of those assets into income streams that could be inserted
into financial instruments. This implies an incessant search for new asset streams ‘which
then—and only then—allows speculation to take place’ (Leyshon and Thrift, 2007: 98).
Although financialization can often imply a world of imaginary financial exchange – abstract
or immaterial forms of capital – financial capitalism is ‘entangled with the ‘‘real’’ economy,
deriving essential flows of value form real assets that constitute the bread and butter of its
reproduction’ (Pani, 2014: 216–217). For Leyshon and Thrift (2007: 98), this requires atten-
tion to the discovery and constitution of:
flows of income that allow all manner of things to companies have to keep a
tether to assets, however far removed, even as they are involved in speculation. The financial
system, like all capitalist entities, must be able to constantly reproduce itself to survive and that
means that it must continuously prospect for new asset seams ...for sustenance and
The language of this analysis is extractive; a prospecting for seams of (latent) material that
could be distilled or converted into objects with financial value. This is an incessant process
in which inert objects are captured and reworked as the basis of financial instruments
(see also Montgomerie, 2008: 245).
Prospecting entails the search for ‘entirely new
income streams ...the identification of a particular geography of revenues which were pre-
viously considered off-limits’ (Leyshon and Thrift, 2007: 101). The kind of extraction which
lies at the mundane beginnings of financialization involves incursion into novel spaces.
‘What we can see now’, they argue, ‘is an impulse to identify almost anything that might
provide a stable source of income, on which more speculation might be built’ (Leyshon and
Thrift, 2007: 98).
This prospecting is an immensely innovative but also mundane practice,
foregrounding the actual techniques with which objects, some quite distant from the world
of finance are converted into financialized objects.
In the rest of this paper, I want to draw on this conception by emphasizing financializa-
tion as a process of assembly, of the ‘making up’, often in mundane terms, of financial assets,
chains and income streams. Rather than conceive, in the first place, of finance as a mode of
accumulation, an orientation to corporate governance, or a mode of culture pervasive across
278 Competition & Change 21(4)
everyday life, I want to highlight a prior process by which objects, practices, spaces and
populations – some distant from or invisible to the mainstream world of finance – are con-
verted into forms that are legible to financial institutions in the first place. This entails what
Muniesa et al. (2017) refer to as a process of capitalization. In their approach, capitalization
refers to the diverse ways that forms of valuation which emphasize return on investment are
constituted and made practical. An object becomes capitalized when, in whatever form, it
becomes legible or calculable as a return on investment. Capital, in this view, is ‘best under-
stood as a process, a relationship, a social relation. Capital not a thing in itself ...but
rather a form of action, a method of control, an act of configuration, an operation’ (Muniesa
et al., 2017: 14).
This approach places emphasis not on finance capital as a structural force
or an already existing object but on the mundane practices – the mechanisms, forms of
knowledge and technologies – which are implicated in the creation of ‘capitalized reality’.
Financialization is a process of valuation, the operation of ‘becoming an asset, becoming an
investment’ (Muniesa et al., 2017: 130).
This ‘becoming an asset’ echoes Leyshon and
Thrift’s concern with the very mundane but basic practices which convert objects and popu-
lations – some quite distant from the formal financial world – into assets or streams of
income (measurable as returns on investment) that can circulate in and through financial
Although the financialization literature has broadened debates regarding the relationship
between finance and society by emphasizing finance as a form of accumulation, a particular
mode of corporate governance and a pervasive factor across everyday cultures, conceiving of
finance as an act of configuration allows attention to the very practical and humble ways in
which financial value is constituted. In the rest of this paper, I want to highlight, in particu-
lar, the importance of visibility in the ways in which forms of financial valuation are
extended into new territories. Muniesa et al.’s characterization of capitalization is littered
with the language and metaphor of the visual, implying that to convert something into a
source of financial value, to allow it to ‘become investment’, is a process that requires
making it visible and subjecting it to a certain line of sight. ‘Capitalization’ Muniesa et al.
(2017: 20) argue ‘is governed by a particular gaze (i.e., a viewpoint, a persona, an outlook,
an angle, a perspective)’.
Visual legibility, however, is deeply reliant on what Bruno Latour describes as inscrip-
tions. Because scientists, programmers and administrators of all kinds cannot directly access
the realities they seek to govern, they must first find ‘inscriptions’ which allow those realities
to be represented in ways that make them apprehensible. Making objects visible is a technical
process involving the ‘paperwork’ – charts, diagrams, maps, calculative devices, ledgers,
tables, graphs – which immerse administrators in a world of ‘hands, eyes, signs’ (Walters,
2002: 91).
Crucial, for Latour, is the conversion of three-dimensional objects into two-
dimensional visualizations. ‘If scientists were looking at nature, at economies, at stars, at
organs’, argues Latour (1986: 15), ‘they would not see anything ...Scientists start seeing
something once they stop looking at nature and look exclusively and obsessively at prints
and flat inscriptions ...[a] simple drift from watching confusing three-dimensional objects,
to inspecting two-dimensional images’
(see also Barry, 2002: 277; Mitchell, 2008: 1120).
For Latour, visible inscriptions are traces of an object which circulate beyond the fixed
locations those objects inhabit. This circulation means that visualizations always operate ‘at
a distance’; are always a removal from or reduction of the realities they attempt to visualize.
A ledger, for example, circulates in spaces far removed from the places where the accounts
were settled or goods actually exchanged. This implies that visual inscriptions have a certain
Aitken 279
kind of presence/absence inherent in the capacity to render something visible and then
transport that visible trace to other, more distant locations. For Latour (1987: 220) this
suggests ‘a whole cycle of accumulation: how to bring things back to a place for someone to
see it for the first time ...How to be familiar with things, people and events, which are
distant’ (emphasis in original). Effective visible inscriptions need to constitute what Latour
describes as ‘immutable mobiles’.
To be ‘present’ somewhere while simultaneously at a
distance requires visualizations which are invariant, which maintain integrity when they
circulate; ‘the many contrivances ...projection, map, log book, etc.—that allow translation
without corruption’ (Latour, 1986: 8). The capacity to manage any object or practice only
occurs when it is constituted in visible form and then removed to other settings in ways that
are not modified in channels of movement.
In a practical sense, this directs attention to the mundane work with which particular
categories and objects are ‘summed up’ and ‘drawn together’. To recover these processes is
to open up critical space around them. The practices of ‘summing up’, notes Latour (1986: 3)
‘are both material and mundane, since they are so practical, so modest, so pervasive, so close
to the hands and the eyes that they escape attention’. These practices, although modest and
close, are also, at the same time, intimately important to the ways in which we are made up
as governable subjects in everyday contexts. To reveal mobilizations is to reveal the mun-
dane practices which make us visible as bodies vulnerable to intervention across a wide range
of settings which often ‘escape attention’.
Taken together, Leyshon and Thrift’s conception of finance capitalism as prospecting and
Latour’s emphasis on inscriptions offer a unique perspective on financialization as ‘acts of
configuration’ which make visible objects as financial assets, as objects that are capable of
being inserted into chains of financial value. In the rest of paper, I want to emphasize
financialization as an attempt to constitute – to configure – capital in and out of the
places where it is now invisible; those bodies which are currently invisible to financial insti-
tutions or which are not legible in the language or forms of display common to mainstream
Seeing like a bank: Making the unbanked visible
Sassen (2014: 1) has recently offered a compelling diagnosis of the global economy as a site
typified by ‘new logics of expulsion ...a sharp growth in the number of people, enterprises,
and places expelled from the core social and economic orders of our time’. This emphasis on
eviction and displacement undoubtedly addresses some of the key and disturbing trends at
the heart of the global economy. It offers, however, an overly narrow account of economic
precarity. Forms of precariousness are not only the result of expulsion but also some pristine
form of exclusion. They also emerge out of complex but often adverse forms of ‘inclusion’
targeted to those at the edges of the global economy. Put a bit differently, precariousness
generated within the global political economy relates not only to forms of expulsion but also
to the terms and conditions with which ‘inclusion’ occurs.
In these terms, global credit practices do constitute a key site of global expulsion.
Globally only 62% of adults have access to any kind of formalized financial account
(Demigruc-Kunt et al., 2015: 2) Recent estimates, for example, suggest that 54 million
Americans and as many as 4.5 billion globally have neither a financial account nor any
form of ‘credit standing’ – constituting a large unbanked population illegible to lenders
(Bornstein, 2014).
This expulsion, however, is deeply entwined with a concerted, but
280 Competition & Change 21(4)
similarly problematic process of incorporation. Unbanked populations are increasingly the
target of dubious programmes of ‘inclusion’; not a simple process of removal, as Sassen
implies, but contradictory processes marked by both absorption and expulsion. Programmes
of ‘financial inclusion’, for example, now proliferate which seek to enfold those outside of
global capital circuits into organized financial markets. During the period between 2011 and
2014, for example, the number of adults globally with access to a formal financial account
increased by 700 million, an achievement often credited to programmes of financial inclusion
(Demigruc-Kunt et al., 2015: 2).
The unbanked appear in these discourses unproblematically as a figure that simply refers
to a lack, a label for an already existing condition. In practice, however, the ‘unbanked’ are
constituted in and through programmes of financial inclusion. In this paper I highlight a
variation of inclusion programming which addresses ‘credit invisibles’; unbanked commu-
nities illegible to lenders. The CFSI (2013) argues that there are 68 million Americans
underserved by the existing credit reporting system, a figure deeply stratified by race.
Racial minorities in the United States, for example, are almost twice as likely as the total
population to be ‘unscorable’ in conventional credit reporting mechanisms
et al., 2015: 6–25; Feinstein, 2013: 4). ‘Credit invisibles’ exist in the crucible of a complicated
cycle: they are unable to access credit because they lack any presence in traditional credit
files which, in turn, is a condition that can only be addressed by taking on, and retiring,
formal credit.
Separating the signal from the noise’: Knowing the unbanked
Those without credit records are neither visible nor knowable to mainstream financial insti-
tutions. To address this, experiments in alternative credit reporting now attempt to make the
unbanked knowable and calculable. ‘Alternative data is essential’, argues FICO, ‘to accur-
ately reflect the financial behaviour and risk of previously unscorable consumers seeking to
join the credit mainstream’ (FICO, 2015: 12; PR Newswire, 2015). These experiments lay
particular emphasis on knowing the unbanked; a contention that ‘a 360-degree view of bor-
rowers has become an even more important component of building a robust loan pipeline’
(Biundo, 2015: 48).
To know the unbanked to ‘know ...customers better by amassing data on a vast scale’
(Wolkowitz and Parker, 2015: 9) – requires the construction of calculative infrastructures
which extract data from ‘alternative’ sites and format that data into forms that are legible to
credit scorers. In Latour’s language this entails the construction of immutable mobiles
visible forms that can move in invariant ways. Calculations are a pristine kind of immutable
mobile, ‘making comparable activities and processes whose physical characteristics and geo-
graphical location are widely dispersed’ (Miller, 2001: 382). Calculation, however, does not
form outside of the conditions that make it possible in the first place. Calculations require an
infrastructure – complex systems of ‘scaffolding’ – which allow measurement:
While calculative infrastructures generate their own relations, they are not themselves automat-
ically generated. They need a way to come into being at all. It requires labor to identify which
conditions provide enough threads, or scaffolding, that could support bringing those various
elements into relation to one another. (Nafus, 2014: 209)
In the rest of this section, I want to outline three particular calculative infrastructures that
are integral to experiments in alternative credit scoring.
The first of these construct fairly
Aitken 281
straightforward calculative mechanisms designed to create immutable assessments of credit-
worthiness based on non-financial payments as proxies for credit history. As early as 2006
advocates began to identify a range of non-credit payment streams – initially energy utilities
and telecommunications services – that could serve as the most ‘credit like’ and ‘most
promising data sets’ (Turner et al., 2006: 11). These payment streams ‘substantially reduce
credit invisibility and enable an estimated 40% of credit invisibles to qualify for some variant
of prime credit’ (Ellison, 2015). The relatively straightforward contention that drives the use
of this broader universe of data is that non-financial payment streams can stand as proxies
for data conventionally used in assessments of creditworthiness.
There have been serious attempts over the past decade to codify this use of alternative
data. Experian, one of the largest global players in the provision of credit analytics, has
helped institutionalize the use of non-financial payments streams via its RentBureau data-
base. RentBureau is a data arrangement which allows rental payments to be systematically
recorded and made available to financial services firms. Rental payment records are of
particular interest to alternative credit scoring experiments because they offer a long and
consistent pattern of payments involving a relatively large proportion of income (Turner
et al., 2015: 12). Consisting of records of over 12 million residents, RentBureau is ‘the largest
and most widely used database of rental payment information ...[and] the first major credit
reporting agency to incorporate positive rental payment data’ (Experian, 2015: 8; see also
Chenven and Schulte, 2015: 7). The system assembles rental data histories by both landlords
who can input on-time payment data as it is received and by renters who can pay their rent
via an online portal that simultaneously captures payment data.
Alongside expansive databases of non-financial payment histories, there have also been
important scoring experiments that use a broader payment data. In 2013, Vantage Score was
launched as a scoring model that combines traditional credit assessment data alongside
rental, utility and mobile phone payment records (Burns, 2013: 16; VantageScore, 2013: 1;
see also Brevoort et al., 2015: 4; Lutz, 2015). A scoring model ‘built with post recession data
and new modelling techniques’ (Burns, 2013: 29) Vantage Score relies on non-financial
payment history for 40% of its weighting making it the most significant formal credit scoring
mechanism that incorporates alternative data (VantageScore, 2013: 1; see also Brevoort
et al., 2015: 4; Lutz, 2015). This attempt to mobilize longer and larger streams of data
into scoring mechanisms has also been facilitated by RiskView, a consumer reporting prod-
uct developed by LexisNexis. RiskView orbits around the contention that ‘consumer life-
cycle and lifestyle data are proven to be highly effective predictors of behavioural patterns
and financial risk levels’ (American Bankers Association, 2016). RiskView integrates a wide
range of payments data into credit scorecards resulting in what it describes as ‘a more
complete picture of consumers’ lives’ (LexisNexis, 2016).
Constituting ‘more complete pictures of consumers’ lives requires calculative infrastruc-
tures which establish regularized ways to gain access to data and to inset those signals into
credit allocation decisions. Much of this trains attention onto technologies which ‘are allow-
ing us to collect, aggregate, and analyze [data] in ways never before possible’ (Baer et al.,
2013: 4). Data access is a key pivot in the calculative infrastructures designed to make use of
non-financial payment streams. Because the data used in these calculations are generated
‘at a distance’ – i.e. not by financial firms but in transactions controlled by other firms –
lenders first need to negotiate points of access which will provide regular payment informa-
tion (Costa et al., 2015: 8). Accessing data for alternative credit scoring arrangements often
282 Competition & Change 21(4)
means negotiating for proprietary or private information, finding ways to address privacy
concerns or using sources of public information.
The calculative infrastructures associated with non-financial payments hinge on data
access. Data, once accessed, becomes a simple proxy for financial markers of creditworthi-
ness. Experiments in alternative credit scoring, however, now range more widely than this
use of non-financial payment streams. A second kind of calculative infrastructure has
emerged which attempts to read creditworthiness by documenting a broader universe of
individual behaviour that can be converted into ‘credit insights’. Credit reporting agencies
now prospect widely for new sources of data in academic records (Zaman and Hawkins,
2015), consumption patterns, public information generated in legal proceedings – bank-
ruptcy, traffic violations, delinquency, tenant–landlord disputes – and, for undocumented
migrants, data relating to remittance payments (CFPB, 2014).
Making visible the unscored is, in part, an attempt to identify traces of stability from this
data. Experian and Equifax have recently launched new measures ‘derived [not only] from
rent, phone, cable, gas and electric payments ...[but also] employment histories and address
changes’ (Kapner, 2012). The assumption that guides much of this interest is that stable
work or residential patterns can be converted into positive risk assessment markers.
Conversely, markers of instability are often translated as signals of unreliability (see
Turner et al., 2015: 18). Put simply, ‘property records that show someone has maintained
the same address for several years the kind of behaviour that indicates someone would
be a good credit risk’ (quoted in Williams, 2015).
Data relating to property ownership,
residential consistency or, conversely, delinquency, itinerancy and ‘irregular’ address
patterns are used as behavioural indicators of creditworthiness over the long term.
At the heart of these new calculative infrastructures are techniques of data ingestion
organized as points of wholesale access to the lives of the credit invisible. Because the
unbanked are more likely to have access to a mobile device than to a formal financial
account, online patterns of digital activity are often used as key points of ingestion. ‘Each
mobile account may generate hundreds or even thousands of calls and text messages per
month, each carrying a rich data set’ (Baer et al., 2013: 3). Mobile phone usage now con-
stitutes one of the largest sources of data about those without regular or formal contact with
global economic institutions (Baer et al., 2013: 5).
As a result, those who are invisible to
credit providers nonetheless often leave distinctly visible ‘digital footprints’ imprinted
through a range of digital practices (Kumar and Muhota, 2012).
Once ingested, data must be sorted by algorithms which categorize mobile users into a
range of behavioural and risk assessment registers (see Bjorkergren and Grissen, 2015).
‘Mobile phone behaviour is not random’, the investor presentation for one key lender notes,
‘it is a rich proxy for consumer lifestyle and choices’ (Cignifi, 2016). One of the most
sophisticated of the calculative infrastructures oriented to online footprints has been con-
structed by Cignifi, an American firm based in Cambridge with ties to the Omidyar Network,
American Express and the International Finance Corporation (IFC). Cignifi does not
directly broker loans but has developed a proprietary algorithm which is used by telecom-
munications firms and their financial partners to make ‘highly predictive risk assessment
decisions about people who have never had a bank account or credit card’ (Schenker, 2015).
Cignifi ingests data from mobile customers related to the number of calls/text messages made
and received per day and the patterns of phone, web and social network usage. It then cali-
brates that data in relation to generic models of behavioural patterns they have developed.
These judgements are used by telecom or financial firms to score unbanked customers.
Aitken 283
For example, Cignifi manages a project in Ghana in collaboration with a local bank,
a telecom provider and the World Savings and Retail Bank Institute. The project ingests
data from customers who are then targeted for financial services. As Cignifi’s (2015a) own
reporting describes it, the unbanked ‘customer base will be segmented using behaviour-based
scores, which will drive targeted marketing campaigns to drive customer acquisition
and usage’.
At the heart of Cignifi’s modelling engines is an analytical link constructed between
mobile usage patterns and behavioural assessment. ‘Mobile usage behaviour opens up a
new data-driven way to qualify ‘unbanked’ for savings products’ (Cignifi, 2016: 17).
Predicting economic conduct – the ability to take on risk, the capacity to bear credit –
can, in this formulation, be read in relation to the degree of immersion in online worlds.
‘Risk’, Cignifi (2016: 19) researchers have concluded, ‘is inversely related to ‘‘connected-
ness’’’; a calculative matrix in which high levels of online connectivity and credit default are
critically related.
Many of Cignifi’s projects extract data from unbanked mobile users in Africa. The data
are reformatted and returned to local telecom, fintech and financial firms who, in turn, use
the assessments to provide credit often extracted from wider circuits of capital. This entails
the extraction of data from local, even intimate settings, and the transit of that data back in
reformatted types of ‘credit insights’. In Uganda, for example, Cignifi (2015b) collaborates
with the IFC and Airtel Uganda, a telecom firm with a wide reach among unbanked cus-
tomers. Cignifi’s (2015b) project promises to access billions of detailed Airtel records relating
to both online footprints and Airtel Money transactions in order to ‘identify prospective
users that have not yet registered for the service’. This kind of calculative infrastructure
constitutes a delicate geography operating at the intersection of personal practice and global
networks of data flow – an infrastructure capable of both ingesting and managing extremely
large streams of data generated from the molecular lives of individuals.
This kind of data ingestion has been replicated, in slightly different ways, by a number of
other experiments, including a series of projects led by First Access, a fin-tech firm based in
New York City. Like Cignifi, First Access extracts data from mobile customers and converts
that data into credit assessments. First Access (2016) ‘combines demographic, geographic,
financial and social data from mobile phones’ in order to develop credit risk scores for the
unbanked. Attempting to ‘separate the signals form the noise’, and to reduce costs for
lenders in ‘informal markets’, First Access has now expanded its app platform across
many parts of the ‘developing world’.
The attempt to know the unbanked through patterns of online or mobile conduct encodes
a complex ambition. At its most basic, this ambition entails a capacity to absorb mass but
raw forms of data about the intimate conduct of daily life. In this sense ‘first access’ has a
kind of double meaning not only as a key innovator in the field but also as an apt description
of a mechanism designed to secure a privileged – first – claim on our online lives in ways that
converts those lives into calculable objects. This ambition also requires identifying relevant
signals within that mass of data and converting those into actionable credit insights. These
processes also rely on a series of legal devices designed to gain lawful access to the data
generated by mobile users (Mazer et al., 2014: 2).
The critical infrastructures that have
resulted stitch together complex scales of operation, often establishing connective tissues
which link the personal and the global; a link between global circuits of credit and deeply
personal spaces echoed in recent attempts to know the ‘psychometric self’.
284 Competition & Change 21(4)
Discovering the psychometric self
The attempt to score those who are beyond the edges of mainstream credit requires
knowing the unbanked in intimate ways. The most ambitious of these involves psychometric
testing, the use of psychological knowledge as a basis for judgement regarding creditworthi-
ness. Psychometric tests are standard assessments designed to determine the mental, behav-
ioural, or in some cases, cognitive capacities of individuals. The use of psychometric testing
in credit practices is premised on an assumption that borrowers ‘have certain types of
personal attributes that can ...predict their future willingness to pay off debts’ (Finberg,
Although initial attempts to measure the psychological profiles of borrowers
entailed the use of paper questionnaires administered mainly by loan officers (see
Figure 1), there has been much recent innovation in the use of mobile apps which extract
psychological data and then convert that data into ‘credit insights’.
This evolution of psychometric testing has been organized, most prominently, by the
Entrepreneurial Finance Lab (EFL) which was founded in 2006 by researchers at the
Harvard Center for International Development. Eventually constituted as a for-profit
firm, the EFL received a founding grant from in 2008 and, by 2010, was helping
to place $1.5 M USD per week. EFL now works with a diverse group of banks internation-
ally and is now active in over 20 countries in Central America and Southern Africa (EFL,
2013: 8–9; McEvoy, 2015: 3). In order ‘to reach applicants without credit history in a
controlled way’, EFL has developed a basic digital tool that leads prospective borrowers
through a 30-minute psychometric test that is administered via a tablet or mobile device
(Multilateral Investment Bank, 2015: 4). The test itself records answers to questions relating
to personality, intelligence, integrity and conscientiousness. The application records not only
the answers to these questions but also a complex web of metadata about how the test was
answered (Meade, 2016). The results are fed into a larger set of algorithms which help
generate scores based on psychological models of creditworthiness.
Kyle Meade (2015), director of Innovation at EFL notes that ‘our data ...create[s] a very
defined outcome: we statistically predict whether a person is likely to default on their loan’.
To do so, EFL has established a distinctive calculative infrastructure designed both to access
Figure 1. Sample Chinese psychometric questionnaire (from Claire, 2013).
Aitken 285
the psychological self and to find ways to translate that into a quantified credit score. This
entails the modelling of patterns which link credit risk with identifiable psychological
characteristics. ‘EFL developed a psychometric credit-scoring tool by first quantifying
the individual characteristics of people who had defaulted on a past loans versus those
who had not’ (Arraiz et al., 2015b: 9). EFL has established a specific grid of behavioural
characteristics it models in relation to creditworthiness: optimism, acumen, self-confidence,
autonomy, opportunism as well as overconfidence (Multilateral Investment Bank, 2015).
In focusing on these kinds of concepts, EFL has slowly placed a particular emphasis on
‘conscientiousness’ as a psychological trait deeply constitutive of the ability to carry credit.
EFL now designs tests as assessments of conscientiousness operationalized as ‘competence,
order, dutifulness, achievement striving, self-discipline and deliberation’ (Claire, 2013: 10;
see also Meade, 2016).
The ambition at the core of psychometric testing is a kind of universality; the possibility
that psychometric testing could be applicable across time and space. In the construction of
‘immutable mobiles’, alternative credit scorers have sought the discovery of psychological
‘constants’ that not only transit without variation but which also are applicable widely.
To address populations with a wide range of educational background, literacy and techno-
logical awareness, EFL designers established a psychological test that relies in particular on
visual appeals. ‘With the use of visual content’, notes its Director of Innovation, ‘EFL has
been able to break down barriers surrounding technology and literacy, thus working towards
our all-inclusive finance mission’ (Meade, 2016).
This emphasis on the visual is echoed by another key innovator in the use of psychometric
testing for credit scoring, VisualDNA, a British fin-tech firm. VisualDNA (2016) confronts
potential borrowers with a visual test (administered digitally) designed to generate a
psychometric score based on five key attributes: ‘openness’, ‘conscientiousness’, ‘extraver-
sion’, ‘agreeableness’ and ‘emotional stability’. VisualDNA has developed a predictive model
designed to assess willingness to repay based on the psychological profile it establishes. This
process is based on what it describes as its unique ‘approach to human understanding’ and
its capacity to link credit risk to the ways in which individuals develop relationships with
others, respond to adversity, are enmeshed in a web of social relationships and have the
capacity to develop positive or optimistic pathways for the future. Because VisualDNA
essentially uses visual cues to determine psychometric scores it has been adapted quickly
in a variety of contexts and is now used in various experiments in credit scoring with retail
banks in Russia, Turkey, Mexico, Malaysia, Poland and South Africa (McEvoy, 2015: 3).
Experiments in alternative credit scoring have taken diverse form, clustered around three
distinctive calculative infrastructures: a web of relationships designed to gain access to non-
financial payment streams, a set of mechanisms capable of ingesting data from social and
online footprints, and a novel series of testing protocols concerned to discover a calculable
and psychometric self. These distinctive calculative infrastructures are nonetheless oriented
around a common rationality – the rendering visible of the unbanked as a body that is legible
to financial institutions. These infrastructures are constituting a ‘field [that] sits at the inter-
section of the explosion of digital data and the rapid development of analytics capable of
mining this data for meaningful insight ...[about those] who are becoming digital and dis-
coverable for the first time’ (Costa et al., 2015: 6). This ‘becoming discoverable’, this latent
visibility, reveals a set of practices ultimately committed to making the unbanked knowable
and calculable.
286 Competition & Change 21(4)
The calculative infrastructures which make the unbanked ‘discoverable’, which convert
it into a visible body, do so by facilitating the flow of data across different scales. This
entails the extraction of data from everyday contexts and the conversion of that into forms
legible to mainstream financial institutions. In covering this distance, these networks suc-
ceed in an ultimate form of immutable but mobile visibility; the conversion of daily prac-
tice into the irreducible and calculative form of a credit score. The unbanked is, however, a
transitory category – a subject characterized simultaneously by the lack which defines it
and the promise that, once visible, it might become completed. Visibility becomes a kind of
conversion; a body that is made visible at the same time that it is rendered something else,
banked and included. This dream of frictionless transit from invisible to visible, from
unbanked to banked, is not matched, however, by a practice of seamless inclusion. In
particular, these attempts to score the unbanked do not uniformly facilitate some whole-
sale incorporation of those outside the mainstream credit system, but rather a set of
dividing practices which sort the bodies capable of bearing credit from those who are
not. Cignifi’s investor presentation (Cignifi, 2016), frames visibility not as a conversion
of raw data into universal access, but as a key to the calculations that might make possible
distinctions in the first place between the ‘credit worthy’ and those who carry ‘high risk’ –
the credit ‘convertible’ and the ‘unprofitable’.
This implies a certain kind of visibility but
also a certain kind of segmentation, a kind of ‘inclusion’ strung in relation to new forms of
social and economic sorting.
‘Making them discoverable’: From inclusion to segmentation
Financialization involves a ‘‘new subjectivity of risk ...installed both for persons and firms, for
policies and movements, for the assessment of value and measure of life. This positive version of
risk is offered all around but is not for everyone. A cleavage runs between the risk-capable and
the at-risk ...’’ (Martin et al., 2008: 129)
The discourse around alternative credit scoring is contradictory. On one hand, these experi-
ments are pitched as a kind of inclusion project, a language which pivots on universal access
and absorption. On the other hand, however, alternative credit scoring is also commonly
associated with a language that is decidedly imperial. The unbanked is approached as an
unknown body in simple need of discovery. The metaphors at work here are ones that
emphasize expeditions which chart the unknown, new territories which can become know-
able, mapped and managed. There is also a decidedly coercive kind of entrepreneurialism
associated with these efforts to score the credit invisible. A language of extraction and brute
removal attends much of the investor presentations, annual reports and marketing material
associated with the alternative credit scoring sector.
The Omidyar Network, for example,
refers to alternative credit scoring as a kind of ‘mining’ of data from the lives that compose it
(Costa et al., 2015: 6). Similarly, VisualDNA consistently frames their credit scoring tests as
a kind of ‘prospecting’ for new borrowers among the risky detritus at the edges of formal
financial system (see VisualDNA 2016).
Prospecting is often framed as a question of natural sensory perception – a need to make
‘them’ visible in ‘our’ line of sight. Despite this appeal to the metaphor of vision and sight,
the actual mechanisms bound up in this making visible are deeply mediated by a range of
human and non-human devices. Alternative credit scores are far removed from any simple
form of human sight and are, rather, reliant on technologies and mediations (data ingestion
mechanisms, algorithmic design, predictive modelling and pattern recognition, machine
Aitken 287
learning of all types) which operate in spaces which are quite distant from human senses. The
trace of these processes that are perceptible to human vision – a final calculative score – is
only made visible, can only be seen at all, after opaque processes of aggregation and
Not seeing anything intelligible is the new normal. Information is passed on as a set of signals
that cannot be picked up by human senses. Contemporary perception is machine to large
degrees. The spectrum of human vision only covers a tiny part of it. Electric charges, radio
waves, light pulses encoded in machines are zipping by at slightly sub-luminal speed. Seeing is
superseded by calculating probabilities. Vision loses importance and is replaced by filtering,
decrypting, and pattern recognition ...a more general human inability to perceive technical
signals unless they are processed and translated accordingly. (Steyerl, 2016)
Producing credit scores for the unbanked involves many, often dispersed, forms of transla-
tion. The calculative outcomes of these countless instances of translation, however, entail an
ambition that is frequently contradictory. As Timothy Mitchell (2002: 119) has rightly noted,
the dreams of calculation are confronted with incoherences which ‘illuminate the impossible
character of calculation’.
In this section I want to emphasize some of the limitations of
alternative credit scoring which will shape its possible futures. The most important set of
limitations relate to the narrow kinds of calculations that alternative credit scoring actually
enables. Although framed as a way to make certain bodies visible, alternative credit scores
often reinforce existing forms of inequality. For some, what is made visible in alternative
credit scoring is not personal characteristics that might allow them to bear risk, but personal
biographies entangled in complex forms of disenfranchisement, historical exclusion or
countless unequal encounters with the systems from which data are drawn. Many of the
sources of alternative credit data are the product of unequal and highly differentiated, even
violent histories. Scores that derive data rental evictions, petty crime records, educational or
academic transcripts neglect the ways in which the conditions these data names are them-
selves entangled in histories of racism and homophobia and emerge out of incalculable but
unequal encounters with all variety or ‘petty sovereigns’ (Butler, 2004): landlords, police
officers, tenant review tribunals, debt collectors and legal professionals. RiskView, for exam-
ple, makes visible ‘transient person attributes’, a negative assessment based on serial address
or employment changes and contact with municipal or landlord/tenant courts (LexisNexis,
2016). Several of the key scoring mechanisms which now make use of alternative data –
FICO Expansion Score, LexisNexis RiskView and L2C Link2Credit – rely centrally on data
from ‘public records’ which include judgements from a wide variety of public legal proceed-
ings: liens, landlord–tenant tribunals, debt settlements and bankruptcies as well as criminal
records (Schneider and Schutte, 2015: 7). Moreover, these conditions, once made visible,
are then given a kind of permanence as calculative certainties. Once particular conditions
or points of experience are reified as calculations they become a reality for those who
experience them; a number without a history or any trace of the inequalities/displacements/
emasculations it encodes. Put simply, if ‘credit scorers rely on non-neutral data collection
tools ...some groups could ultimately be treated less favorably or ignored’ (Hurley and
Adebayo, 2016: 178). This is not to suggest that there are no tangible benefits which
accrue to those who become credit visible – research now suggests that alternative credit
scoring frequently reduces the cost of credit to some of those it addresses (Experian, 2014: 8;
see also Ellison, 2015).
This is to suggest, however, that for many others alternative
credit scoring can make visible conditions – evictions, criminal records, racialized
288 Competition & Change 21(4)
convictions for petty crime – which can cement exclusion from, rather than access to cheaper
forms of credit.
Appleyard et al. (2016a: 300) make a compelling suggestion to address issues of ‘varie-
gation’ and ‘the uneven ways in which financialisation plays out in practice over space’.
Although experiments in alternative credit scoring are relatively new, there is evidence
accumulating that in some cases credit scores derived from alternative data and algorithmic
machine learning may lead to new kinds of variegation; a risk that the credit invisible may be
better off without a credit score that can expose them to deepening forms of economic or
financial disenfranchisement. As Wu (2015) puts it ‘while credit invisibility poses real and
significant problems some areas, no credit history is better than a bad one’. In part this
is an acknowledgement that some of the data that are harnessed in alternative credit scores
can lead to disproportionately negative assessments of creditworthiness. Alternative credit
scores by design take into account data not considered within mainstream credit networks.
This includes source data which exert stricter standards of creditworthiness than those which
predominate in mainstream credit networks. Experiments which take into account repay-
ment data for payday loans, for example, encode and reflect the kinds of serialized debt
pressures which disproportionately enmesh poor populations in ongoing cycles of debt accu-
mulation (see Aitken, 2015). Similarly, new innovations that allow the absorption of rental
payment data often include full monthly results which categorize borrowers ‘in arrears’ after
only 30 or 60 days, a much more stringent standard than those common within mainstream
or prime credit markets (Wu, 2015).
Evidence is also accumulating that alternative credit scores are nestled within and repro-
duce existing racial inequalities.
Many of the practices and institutions out of which the
raw data for alternative credit scores are generated, are themselves the product of deeply
racialized processes. Onerous judgements from debt collection lawsuits, for example, are
disproportionately more likely among minority than white communities.
that absorb data generated from racialized sources encode and carry those very differences
in the calculative measures they generate (Yu and McLaughlin, 2014: 27).
Moreover, machine learning credit scores racialize credit markets in novel kinds of ways.
Of particular importance are the ways in which data related to social media clusters or
consumption patterns become proxies for particular communities. Many alternative credit
scoring models gather geographical information which is then correlated with patterns of
credit risk. Credit data, for example, are often extracted and coded in relation to IP address,
zip code or census information all of which can indicate geographical parameters for assess-
ments of creditworthiness (Hurley and Adebayo, 2016: 182). These geographical factors, by
extension, take into account race in ways that echo the longer history of minority neigh-
bourhoods redlined and constituted as mortgage risky. Location, argue Yu and McLaughlin
(2014: 27) succinctly ‘can function as a proxy for race and income’. Although the gathering
of information regarding race in credit assessment is prohibited by federal law, machine
learning models place individuals in ways that are racialized, that can read creditworthiness
in relation to race. Moreover, once these patterns are established in particular sets of algo-
rithms, they become increasingly important variables. If in the the ‘process of machine
learning, the model learns that race highly correlated to credit risk, the model will
attach greater significance to proxy variables that can serve as a stand-in’ (Hurley and
Adebayo, 2016: 182–183). As numerous commentators have now noted, and as credit scorers
have explicitly begun to formulate, credit assessment often requires placing credit invisible
populations in ways that allow them to be easily coded in relation to the spaces and cultures
Aitken 289
they occupy. Locating credit invisibles within social networks or particular physical geogra-
phies allows credit assessments which are determined, at least in part, by a reading of those
geographies as either risk capable or simply risky. As a recent TransUnion report puts it
‘aggregated credit data ...can identify local credit conditions clustered around common
demographics. This is especially true for consumers with little or no credit history’
(quoted in Yu and McLaughlin, 2014: 28).
Alternative credit scores do not assess credit in relation to individualized traits directly
observed, but in relation to quantifiable and regular patterns gleaned from the assessment of
increasingly large sets of big data. This can result in low credit scores not well coordinated to
particular circumstances or individualized conditions. These problems are exacerbated
because scores based on machine learning and big data frequently entail a codified rigidity.
Some data streamed into algorithms are inaccurate or imprecise. Because algorithms
are fed from increasingly large and diverse data points they are vulnerable to errors and
inconsistencies. Moreover, data that are used in the creation of alternative credit scores
are proprietary and, as a result, are insulated from scrutiny and transparency. Alternative
credit data ‘tools are non-transparent and rely on inaccurate data collected form numerous
sources, making it difficult for consumers to verify or challenge unfair decisions’
(Hurley and Adebayo, 2016: 166). At the heart of American consumer credit protection
schemes is the right of consumers to challenge the ‘accuracy or completeness’ of information
contained in individual credit files – a principle of transparency inapplicable to proprietary
mechanisms that subject large streams of data to machine learning outcomes (Yu and
McLaughlin, 2014: 24).
Taken together these factors have caused some alarm that alternative credit scoring could,
at least in some contexts, simply convert the credit invisible into populations designated as
prohibitively risky. Making the financially excluded visible not as potential credit bearers,
but as bodies marked as dangerously risky has wide implications. Wu (2015) argues that the
codification of low credit scores not only make it more costly for the unbanked to access
financial services but can also lead to broader and deepening forms of economic precarity.
Newly constituted visibility in the form of a negative credit judgement can undermine job
prospects and expose borrowers to predatory lenders who often target vulnerable, desperate
or low-score customers. This in turn exposes vulnerable populations to more onerous eco-
nomic burdens and costs. As Alloway (2015) has succinctly put it, using alternative big data
and machine learning to make the unbanked visible can actually create a negative sell-
fulfilling cycle, ‘hardening the lines between the wealthy and the poor by denying credit to
those who are already associated with not having access to it’. The variegation at the heart of
this particular kind of financialization is one that risks deepening rather than constricting the
complexities of financial ‘inclusion’ and ‘exclusion’.
As Martin suggests, financialization is a process deeply implicated in the ‘cleavages’ that
separate the ‘risk capable’ and the ‘at risk’. As a particular kind of financialization oriented
to the ‘credit invisible’ – to finding ways of making them visible within chains of financial
value – alternative credit scoring experiments produce what Sassen (2014: 4) has referred to
as ‘a savage sorting’. In contrast to the ways in which it is often imagined as a pristine form
of inclusion, alternative credit scoring is more aptly delineated as a kind of social sorting, a
mechanism designed to sort and distinguish the bodies that could render profit from those
that might impair it. In these terms, and although alternative credit scoring is often framed
as an experiment in risk bearing, in the expansion of ways in which the unbanked might be
encouraged to confront and bear risk, it can often be, in practice, a technique of risk
290 Competition & Change 21(4)
aversion, a practice in which credit providers find profitable ways to manage and avoid
bodies they perceive as risky.
In this risk-aversion strategy, lenders simultaneously want to extend credit to the
unbanked – what Lutz (2015) refers to as the need to ‘expand the loan pool’ – but are
also concerned to minimize their exposure to risk in doing so. In a foundational way,
alternative credit scoring techniques manage this tension by segmenting those who are
unbanked but low risk from those who are irredeemable default risks; experiments premised
on an assumption ‘that there are sizeable shares of the credit invisible that are of low and
moderate risk’ (Turner et al., 2015: 17). These experiments are designed not so much to make
the unbanked visible but to segment the unbanked into those who are potential borrowers
and those who are not. The EFL (2013: 6) tool, for example, is described explicitly as a
mechanism that allows for ‘risk sorting on difficult to evaluate segments—unbanked, inde-
pendent workers ...thin file’ borrowers.
This segmentation, however, is explicitly designed to extract financial value and profit
from the unbanked. Experian (2011: 12), for example, delineates its approach as a search for
the ‘segments of this population with the most manageable risk’.
This prospecting and
identification of the risk-manageable segments of unbanked populations is itself a
key to operational efficiency and, by extension to value derived from lending practices.
One credit analytics firm, for example, estimates that alternative credit scoring models
allow lenders to reduce costs associated with loans in the subprime sector by 20–50%
(Baer et al., 2013: 2). In general, profitable segmentation allows for ‘better risk discrimin-
ation, better operational efficiency and pricing’ (Alliance for Financial Inclusion, 2014).
Because credit scores for the unbanked allow for both an expansion of the loan pool as
well as a careful segmentation of that pool in ways that avoids risk, they are, in ultimate
form, a method for deepening the financial value of loan portfolios held by lenders.
International financial institutions such as the World Bank have consistently promoted
the EFL model as a ‘tool [which] can be used to make additional loans ...without increasing
the bank’s portfolio risk’ (Multilateral Investment Bank, 2015: 4).
Taken together, these
ways of approaching the unbanked are centred on the extraction of value in a stabilized and
managed way from a population considered risky. Alternative credit scoring becomes, in this
formulation, a ‘skimming mechanism for applicants rejected under the traditional credit-
scoring method, [and a way] to offer more loans without increasing the risk of the portfolio’
(Arraiz et al., 2015a: 3; see also Arraiz et al., 2015b: 4). The unbanked are discovered in these
experiments not as bodies designated for inclusion in any simple way, but as financialized
objects in their own right, as sources of financial value that can be measured, ‘skimmed’, and
Conclusion: Keeping the outsiders out
This paper focuses on experiments in alternative credit reporting as a site key to the
unbanked and to the ways in which the unbanked is constituted as a category of knowledge
and intervention. Although it seemingly refers to an unproblematic category – the residue
left outside of global political–economic practices – the unbanked is an object that is pro-
duced in the practices that are said merely to know it. Experiments in alternative credit
scoring are one site where the unbanked are rendered knowable and visible in particular
kinds of ways. This entails a particular kind of financialization in which financial value is
constituted in and extracted from the places where it is now invisible. To make visible is also,
Aitken 291
at the same time, the capacity to make governable. As I have argued in this paper, alternative
credit scoring, and its invocation of complex machine learning models, render the unbanked
governable in two opposing ways. Some unbanked are made up in these experiments in a
transitory way, as a population that could be remade as they are absorbed into formalized
credit networks and into the chains of valuation those networks generate. For other seg-
ments of the unbanked, however, these experiments make them visible as bodies that are too
risky to carry value. If financialization is an ‘act of configuration’ it is one that can make
visible both new forms of valuation as well as people or places devoid of value.
For Frances Stonor Saunders (2016: 8), the translation of signals into forms of big data is
ultimately bound up in a ‘a verified self, an identity, formed through and confirmed by iden-
tification, that is attested to be true’. This ‘verified self’ signals a preoccupation with bodies
that could be known with certainty and completeness. The ambition of perfect verification
fuels the proliferation of digital and biometric mechanisms designed to extract data from all
variety of personal patterns, behaviours, movements and physiological characteristics. In this
pursuit, the body becomes itself a kind of ‘database from which some sort of content is
extracted or ‘‘captured’’, then algorithmically encrypted and sorted for retrieval’ (Saunders,
2016: 8).
Alternative credit scoring is a particular variation of this verified self. Experiments
in alternative credit scoring are, in some essential measure, attempts to know the unbanked –
to know unbanked bodies, payment traces, psychological inclinations, online behaviour, social
footprints – and to verify the creditworthiness of those bodies in detailed and intimate ways.
For Saunders, however, appeals to the verified self unleash dangerous dividing practices.
The ambition to know and verify bodies marks a kind of ‘medieval modernism ...born of a
fatal resolve to keep the outsider out, to separate the verified from the unverified’ (Saunders,
2016: 11). Verification is a relational move which specifies in granular form those bodies that
can be certified and those that are simply beyond the reach of contemporary regimes of
bureaucratic or managerial visibility. Experiments designed to make the unbanked visible,
although framed as a kind of simple gesture of universal inclusion, are actually a complex
practice of segmentation; a site where the kinds of bodies who can bear credit are calculated
and where the lines between them and those irredeemably outside of formal credit practices
are sharpened. Building on but also inverting Sassen’s argument, alternative credit scoring/
verification is not a site of either inclusion or expulsion but of the complex relations between
the two.
In this sense, the ambitions for a verified self confront those on the edges with a cruel
paradox. On one hand, to be verified, to be rendered visible is often the only conceivable
channel for those seeking to move beyond their own precarity. Visibility within mainstream
credit practices can, at least in some contexts, facilitate access to vital financial resources.
On the other hand, however, for those at the edges of formal credit practices, verification and
visualization in practice often result in segmentation, entrenched forms of credit denial, or a
further deepening constriction of options. For some, verification can only mean exclusion
confirmed. To have an unscorable or irredeemable credit status verified is to shrink the space
available and to reify that shrinkage into calculable form. ‘Our system of verification’, argues
Saunders (2016: 11), is only ‘a mechanism for sending them back’. It is in the crucible of
these pressures – verification as access for some but as removal for others – that the dream of
visibility is revealed as an ambition that links in inextricable ways the elegant fictions of
counting with the prosaic but harsh lines which determine who actually counts; which bodies
actually count as something and which ones are beyond reach.
292 Competition & Change 21(4)
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or
publication of this article.
The author(s) received no financial support for the research, authorship, and/or publication of this
1. This argument draws, in important ways, on an argument developed by William Walters (2010: 73)
on the production of the category of ‘illegal immigration’.
2. Martin reiterates this elsewhere (Martin, 2006): ‘The twin formations of contemporary
finance—securitization and derivatives—are more saliently principles of movement, the first for-
cing association or assembly out of disparate risks, debts and productive capacities, the second
disassembling and dispersing established equities or entities’.
3. Much of this following paragraph is a revised and reworked version of Aitken (2015: 37–38).
4. Watkins (2017: 99) offers a similar general view by suggesting that ‘financialization refers to the
increasing power of financial institutions rising income that accrues to financial institutions,
in changing policies that benefit financial institutions, and in financial innovations to develop new
income streams’.
5. As Epstein (2005: 3) has famously put it, financialization ‘means the increasing role of financial
motives, financial markets, financial actors and financial institutions in the operation of the
domestic and international economies’.
6. See also Baragar and Chernomas (2012: 337), Blackburn (2006: 43–44), Christopherson et al.
(2013: 351), Dembinski (2009), Krippner (2005: 181) and Van der Zwan (2014: 104).
7. It is important to note that for some observers, financialization is less a material process than a
form of knowledge. For Knorr Cetina and Preda (2007: 116–117) finance is ‘generated entirely in a
symbolic space—the market world is informational’.
8. As Blackburn (2006: 42–43) argues, ‘making a good profit is no longer enough; a triple A rating is
also needed’.
9. Put succinctly, ‘financial markets exert pressures on non-financial corporations, and the man-
agers running them, to adopt business practices promoting shareholder value’ (Van der Zwan,
2014: 107).
10. For a useful assessment of the importance of knowledge/power in relation to financialization,
see Montogmerie (2008: 243): ‘Financialization offers an account of the systematic ‘‘knowledge’’
performed in markets which has a discursive power all of its own’.
11. In Paul Langley’s (2008: 134) terms, finance is ‘constantly assembled and re-assembled through
processes of identitification that extend through the centres of ‘‘high finance’’ and the corporate
boardroom to the workplace, shopping mall and home’.
12. For Pathak (2014: 91) this has resulted in a ‘normalization of indebtedness and the stigmatization
of over indebtedness as a corollary of ‘‘delinquent’’ dispositions and dependencies’.
13. These kinds of variegated analyses provide a ‘missing link between economic discourse (and finan-
cial instruments) and everyday practice’ by inserting themselves into the distance between ‘what
finance invites people to do ...[and] what they actually do in their daily lives’ (Pellandini-Simanyi
et al., 2015: 736; see also Appleyard et al., 2016b: 530). Appleyard et al. (2016b: 530) argue
that financialization of everyday culture has focused usefully ‘on issues of culture, identities and
subjectivities ...but ...fails to fully engage with the ‘‘lived experience’’ or ‘‘lived reality’’ of
Aitken 293
financialisation’. The work of Fernandez and Aalbers (2016: 72) is particularly striking in this
respect – an argument that places housing at the heart of the most recent wave of financialization:
the abosrption of capital by the housing sector and real estate more generally was one of the
defining characteristics of the current age of financialization, exceptionally inflating the
balance sheet of households and banks ...housing was the main collateral for the debt-
driven process of financialization.
14. I am grateful to Randall Germain for his observation regarding the ways in which financialization
addresses itself to hitherto inert objects.
15. Leyshon Thrift (2007: 99) affirm that ‘it is now possible to use just about anything as a platform
for more speculative financial activity ...just so long as it represents a stable and continuing
income stream’.
16. They note that ‘capitalization is an event, an empirical moment, an occurrence, an action ...Close
examination of the scenery, the narratives that are deployed ...becomes a critical component of
this inquiry’ (Muniesa et al., 2017: 17).
17. This approach seeks to overcome any simple divide between ‘real’ and ‘fictitious’ forms of capital.
‘We focus on valuation as an operation: an operation that is real as soon as it takes place, an
operation that produces reality as soon as it has effects’ (Muniesa et al., 2017: 15).
18. This focus on inscriptions recasts ‘representation as an ‘‘active, technical process’’; less a question
of systems of meaning, more a matter of its techne...’ (Walters, 2002: 91).
19. As Latour (1986: 20) argues ‘everything, no matter where it comes from, can be converted into
diagrams and numbers, and combination of numbers and tables can be used which are still easier
to handle than words or silhouettes’.
20. An immutable mobile is, put simply, ‘a device that makes both mobilization and immutability
possible at the same time’ (Latour, 1986: 10).
21. In part this analysis is influenced by James Scott who has argued, the exercise of power is deeply
implicated in forms of visual legibility. The sovereign state, according to Scott, is unable to assert
authority without finding mechanisms – cadastral maps, uniform systems of weights and measures,
censuses, plans of all kinds – with which to make visible the population and territory which is to be
governed. ‘Legibility’, Scott (1998: 76–78) suggests, ‘implies a viewer whose place is central and
whose vision is synoptic’. Although it is a bit beyond the scope of this paper, note that I am
drawing on Latour’s more decentred conception of the power of visualization rather than Scott’s
conception of single viewer which emphasizes an overly centred conception of sovereign power.
‘Talking of power,’ notes Latour (1986: 27) ‘is an endless and mystical task; talking of distance,
gathering, fidelity, summing up, transmission, etc., is an empirical one’.
22. Some reports suggest that the figure of unbanked Americans is actually 64 million. See, for
example, Connor (2014: 3).
23. ‘...on average, blacks and Hispanics have lower credit scores than non-Hispanic whites and
Asians...’ (Federal Reserve, 2007: 8).
24. Brevoort et al. (2015: 24) estimate that there are 26 million adults in US without a credit record
alongside another 19 million with records that are ‘unscorable’. The Consumer Financial
Protection Bureau reports that approximately 15% of Blacks and Hispanics are ‘credit invisible’
compared to 9% for the general population (Brevoort et al., 2015: 6–25).
25. In the words of one advocate, alternative credit reporting requires ‘the ability to process and
analyze personal data has the potential to unlock tremendous insight into consumers around
the world’ (Costa et al., 2015: 20).
26. This is a point that is echoed by others who point out that data itself must first be created. ‘At first
glance data are apparently before the fact: they are the starting point for what we know, who we
are and how we communicate...[but] ‘Data [do] not just exist...they have to be ‘‘generated’’. Data
need to be imagined as data to exist and function as such’ (Gitelman and Jackson, 2013: 2–3
emphasis in original).
294 Competition & Change 21(4)
27. Recent research by TransUnion shows that there has both been growth in the adoption of alter-
native data in credit scoring but it has been somewhat uneven. Alternative data are most frequently
used by credit card lenders and consumer finance firms (over 50% of which now incorporate
alternative data in some form while 36% of banks and 16% of credit unions do (TransUnion,
2016: 15–16).
28. The inclusion of a broader range of payments streams also makes visible the ability to consummate
regular payments over time in ways that would not be visible in conventional credit scoring prac-
tices. Incorporating non-financial payment data into credit files also rests on the assumption that
‘more information on the consumer credit file enables a better assessment of the consumer’s risk
level’ (Experian, 2014: 5; VantageScore, 2015: 2).
29. A similar systematic database has been assembled by Equifax, one of the ‘big three’ American
credit reporting agencies. The Equifax Consumer Services Database contains both positive and
negative payment data for 171 million consumers, 25 million of which have no traditional credit
file (CFSI, 2013).
30. Access includes issues of ‘regulatory compliance, depth of information, scope of coverage, accur-
acy, predictiveness and orthogonality’ (Turner et al., 2015: 14).
31. This ‘universe of information’, includes ‘patterns of purchase, measures of intent, inferences about
character, schools attended, degrees earned, how you used your education, your social circle, your
spelling and more’ (Browdie, 2015).
32. Or, as put by FICO (2015: 8), ‘consumers who have been at their address for a longer period of
time are more likely to pay their credit obligations than those more transient’.
33. This is key to RiskView, a scoring mechanism oriented to the search for stable economic
lives among the noise of big data. RiskView places emphasis on what it describes as markers of
individual in/consistency: ‘education attributes’, ‘characteristics of previous address attributes’,
‘transient person attributes’ ‘phone and higher risk address attributes’ (LexisNexis, 2016; Tewari
and Shellenberger, 2014: 14).
34. ‘Mobile phone data is the most ubiquitous form of data in emerging markets’ (Cignifi, 2016: 24).
35. The basic contention of these experiments is that ‘behavioural signatures in mobile phone data
predict default with accuracy approaching that of credit scoring methods that rely on financial
histories’ (Bjorkergren and Grissen, 2015).
36. This approach is also echoed by Lenddo (2016). Lenddo’s (2014) patent application describes its
own practice as a ‘solution for ...determining a risk-related score ...determined based on avail-
able personal data gathered from the online social footprint ...generated by applying a predictive
model to the data’.
37. Firms have been preoccupied with the innovation of ‘methods for informing borrowers
their data will—and will not—be used’ (Mazer et al., 2014: 2).
38. The Alliance for Financial Inclusion (2014), for example, reports that basic psychometric testing
involves intelligence, business skills and a suite of issues relating to confidence, attitudes and
39. Psychometric tests assess a diverse range of conduct: ‘those who add last names to their contacts or
those who are known gamblers ...are likely to be more creditworthy. On the other hand, those
who drain their battery quickly or travel infrequently are likely to be less creditworthy’ (Samlin,
40. Cignifi’s presentation (2016) includes an image that links the analysis of raw data into ‘valuable
consumer insights’ that include the categorization of ‘millions of customers’ into four categories:
‘credit worthy’, ‘high risk’, ‘low conversion’ and ‘unprofitable’.
41. See Steyerl (2016) who uses this extractive language in general to describe big data: ‘Today,
expressions of life as reflected in data trails become a framable, harvestable, mineable resource
managed by informational biopolitics’.
42. These experiments place particular emphasis on prospecting as a process of using ‘data from
multiple sources accurately identify individual needs, better know their customers, and
Aitken 295
help customers better know their own habits and financial patterns’ (Wolkowitz and Parker,
2015: 6).
43. Mitchell’s argument also implies the need to also pay attention to the question of resistance.
Although it is beyond the realm of this paper, it is important to pay attention to populations
‘who have developed their own notions of analytics that are separate from, and in relation to,
dominant practices of firms and institutionalized scientific production ...critical sense-makers’
(Nafus and Sherman, 2014: 1785).
44. As Ellison (2015) attests, in the migration ‘from thin file to thick file ...[borrowers] received
modelled interest rates more than two percentage points lower’.
45. Although for some ‘having a low score is better than no score’ (Ellison, 2015) for others having a
credit score becomes an exercise in which existing expulsions are reified in calculative form and
become the basis of further distance from affordable credit. The implication of this analysis is that
‘inclusion’ is not always politically viable or desirable. Although it is beyond the scope of this
paper, the political connotations are important. What needs to be worked out in greater detail is
how and in what political conditions critical scholarship would promote ‘invisibility’ rather than
visible legibility. ‘Illegibility, then, has been and remains a reliable resource for political autonomy’
(Scott, 1998: 54).
46. For a general sense of the debate, see White (2017).
47. Wu (2015) makes the general argument: ‘credit scoring is a reflection of the racial economic divide
in this country ...income and wealth disparities are caused by centuries of discrimination....Credit
scoring perpetuates this unequal playing field’.
48. This insight comes from a Problica study referenced in National Consumer Law Center
(2016: 1).
49. The language used by key players in the alternative credit scoring networks is preoccupied with risk
and its management (FICO, 2015: 5–6).
50. Put a bit differently, Experian (2011: 14) notes that a key to alternative credit scoring is a capacity
‘to target viable niches within the subprime population...[and] allow lenders to zero in on the most
responsive segments of prospects’.
51. This, in turn is ultimately aimed at developing strategies that might carefully ‘generate and accel-
erate portfolio growth’ (Experian, 2011: 10).
52. As Saunders (2016: 9) argues, ‘the verified self becomes a perpetual confession in which your body
and mind are testifying for you without you even realizing it’.
Aalbers M (2008) The financialization of home and the mortgage market crisis. Competition and
Change 12(2): 148–166.
Aalbers M (2016) The Financialization of Housing: A Political Economy Approach. London: Routledge.
Aitken R (2015) Fringe Finance: Crossing and Contesting the Borders of Global Capital. London:
Alliance for Financial Inclusion (2014) Financial Innovation with Big Data Analytics. Kuala Lumpur:
Alliance for Financial Inclusion.
Alloway T (2015) Big Data, Finance and Inequality. Available at:
s¼finance (accessed 12 April 2017).
American Bankers Association (2016) Alternative Data Credit Risk Decisioning Solution—LexisNexis.
New York: American Bankers Association.
Appadurai A (2016) Banking on Words: The Failure of Language in the Age of Derivative Finance.
Chicago, IL: University of Chicago Press.
Appleyard L, Rowlingson K and Gardner J (2016a) The variegated financialization of sub-prime credit
markets. Competition and Change 20(5): 297–313.
296 Competition & Change 21(4)
Appleyard L, Rowlingson K and . Gardner J (2016b) Payday lending in the UK: the regul(aris)ation of
a necessary evil? Journal of Social Policy 45(3): 527–543.
Arraiz I, Bruhn M and Stucchi R (2015a) Psychometrics as a Tool to Improve Screening and
Access to Credit. Washington, DC: Inter-American Development Bank/Multilateral Investment
Arraiz I, Bruhn M and Stucchi R (2015b) Psychometric Tests as a Tool to Improve Screening and Access
to Credit: Evidence from Peru. Washington, DC: Inter-American Development Bank.
Baer T, Gorland T and Schiff R (2013) New Credit-Risk Models for the Unbanked. New York:
McKinsey and Company.
Barry A (2002) The anti-political economy. Economy and Society 31(2): 268–284.
Baragar F and Chernomas R (2012) Profits from production and profits from exchange:
Financialization, household debt and profitability in 21st century capitalism. Science & Society
76(3): 319–339.
Biundo R (2015) Looking beyond credit scores. Mortgage Banking, 1 August 2015, 46–55.
Bjorkergren D and Grissen D (2015) Behaviour Revealed in Mobile Phone Usage Predicts Loan
Repayment. Unpublished manuscript.
Blackburn R (2006) Finance and the fourth dimension. New Left Review 39: 39–70.
Bornstein D (2014) ‘Invisible’ Credit?: (Read this Now!). New York Times, 2 October.
Brevoort K, Grimm P and Kambara M (2015) Data Point: Credit Invisibles. Washington, DC:
Consumer Financial Protection Bureau/CFPB Office of Research.
Brighenti AM (2010) Visibility in Social Theory and Social Research. New York: Palgrave.
Browdie B (2015) Can alternative data determine a borrower’s ability to repay?. American Banker
180(309): 1–2.
Burns B (2013) Expand and Grow: How to Reach and Education More Members Using a New Breed of
Credit Scores. New York: VantageScore Solutions LLC.
Butler J (2004) Precarious Life. London: Verso.
CFPB (2014) Report on the Use of Remittance Histories in Credit Scoring. Washington, DC: Consumer
Financial Protection Bureau.
CFSI (2013) Uncovering Opportunity: The True Potential of the Underserved Market. Chicago, IL:
Center for Financial Services Innovation, Webinar.
Chenven S and Schulte C (2015) The Power of Rent Reporting Pilot: A Credit Building Strategy.
New York: Credit Builders Alliance/Citi Foundation.
Christopherson S, Martin R and Pollard J (2013) Financialisation: Roots and repercussions.
Cambridge Journal of Regions, Economy and Society 6: 351–357.
Cignifi (2015a) Airtel Partners Cignifi and WSBI to Develop Innovative Mobile Savings Product.
Cambridge, MA: Cignifi.
Cignifi (2015b) Cignifi and the IFC to Grow Usage of Mobile Money in Uganda. Cambridge, MA:
Cignifi (2016) Using Mobile Data to Reach the Unbanked: Investor Presentation. Cambridge, MA:
Claire D (2013) Using ‘Psychometric’ Factors to Improve Credit Scoring Models for Small Business
Lending. Edinburgh: University of Edinburgh Business School.
Connor N (2014) Deposit Accounts: Alternative Data Offers Fresh Strategy and Big Opportunities.
New York: Equifax.
Corkery M and Silver-Greenberg J (2015) Banks reject New York City IDs, leaving ‘unbanked’ on
sidelines. New York Times, 23 December.
Costa A, Deb A and Kubzanksy M (2015) Big Data, Small Credit: The Digital Revolution and its
Impact on Emerging Market Consumers. San Francisco, CA: Omidyar Network.
Davies M (2012) The aesthetics of the financial crisis: Work, culture and politics. Alternatives: Global,
Local, Political 37(4): 317–330.
Aitken 297
Davis A and Walsh C (2016) The role of the state in the financialization of the UK economy. Political
Studies 64(3): 666–682.
Davis L (2016) Identifying the ‘financialization’ of the nonfinancial corporation in the U.S.
economy: A decomposition of firm-level balance sheets. Journal of Post Keynesian Economics
39(1): 115–141.
Dembinski P (2009) Finance: Servant or deceiver. Financialization at the Crossroad. New York:
Observatoire de la Finance/Palgrave Macmillan.
Demigruc-Kunt A, Klapper L, Singer D, et al. (2015) The Global Findex Database 2014: Measuring
Financial Inclusion Around the World. Washington, DC: World Bank Group, Policy Research
Working Paper 7255.
EFL (2013) Credit Bureau Data Reliable Scoring Presentation. Cambridge: EFL Global Ltd.
Ellison K (2015) Five ways alternative data can expand credit access. American Banker 180(95).
Epstein G (2005) Financialization and the World Economy. Cheltenham: Edward Elgar.
Experian (2011) Expanding the Marketable Universe. New York: Experian Information Solutions Inc.
Experian (2014) Let There Be Light: The Impact of Positive Energy-Utility Reporting on Consumers.
New York: Experian Information Solutions Inc.
Experian (2015) Credit for Renting: The Impact of Positive Rent Reporting on Subsidized Housing
Residents. New York: Experian Information Solutions Inc.
Federal Reserve (2007) Report to the Congress on Credit Scoring and its Effects on the Availability and
Affordability of Credit. Washington, DC: Board of Governors of the Federal Reserve System.
Feinstein J (2013) Alternative Data and Fair Lending. New York: LexisNexis Risk Solutions Bureau.
Fernandez R and Aalbers M (2016) Financialization and housing: Between globalization and varieties
of capitalism. Competition and Change 20(2): 71–88.
FICO (2015) Can Alternative Data Expand Credit Access?: Insights White Paper No. 90. New York:
Fair Isaac Corporation.
Finberg R (2016) Fintech spotlight: Data vs. data—The battle to understand online lending borrowers.
Finance Magnates, 3 March.
First Access (2016) Instant Risk Scores for Emerging Markets. New York: First Access Inc.
Gitelman L and Jackson V (2013) Introduction. In: Gitelman L (ed.) ‘Raw Data’ Is an Oxymoron.
Cambridge: MIT Press, pp.1–14.
Harmes A (2001) Mass investment culture. New Left Review 9: 103–124.
Hiss S (2013) The politics of the financialization of sustainability. Competition and Change 17(3): 234–
Hurely M and Adebayo J (2016) Credit scoring in the era of big data. The Yale Journal of Law and
Technology 18: 148–216.
Kapner S (2012) For ‘credit invisibles,’ a market takes shape—Firms devising new methods to score the
previously shut-out. Wall Street Journal, 20 November, C.1.
Knorr Cetina K and Preda A (2007) The temporalization of financial markets: From network to flow.
Theory, Culture and Society 27(7/8): 116–138.
Krippner G (2005) The financialization of the American Economy. Socio-Economic Review 3(2): 173–208.
Krippner G (2011) Capitalizing on Crisis: The Political Origins of the Rise of Finance. Cambridge, MA:
Harvard University Press.
Kumar K and Muhota K (2012) Can Digital Footprints Lead to Greater Financial Inclusion?
Washington, DC: Consultative Group to Assist the Poor.
Langley P (2008) Financialization and the consumer credit boom. Competition and Change 12(2): 133–147.
Latour B (1986) Visualization and cognition: Thinking with eyes and hands. In: Kuklick H (ed)
Knowledge and Society Studies in the Sociology of Culture Past and Present. Greenwich CT: Jai
Press, pp.1–40.
Latour B (1987) Science in Action: How to Follow Scientists and Engineers Through Society.
Cambridge, MA: Harvard University Press.
298 Competition & Change 21(4)
Lenddo Limited (2014) A Risk-Related Scoring Using Online Social Footprint. Washington, DC:
United States Patent and Trademark Office, Patent application no. 14/180622.
LexisNexis (2016) Alternative Data Credit Risk Decisioning Solution–Lexis Nexis Risk View. American
Bankers Association.
Lenddo Limited (2016) LenndoScore. New York: Lenddo Limited.
Leyshon A and Thrift N (2007) The capitalization of almost everything: The future of finance and
capitalism. Theory, Culture and Society 24(7–8): 97–115.
Li TM (2014) ‘‘What is land?’’: Assembling a resource for global investment. Transactions of the
Institute of British Geographers 39: 589–602.
Lutz H (2015) New credit scoring deepens loan pool. Automotive News 89(6702).
McEvoy M (2015) Enabling Financial Inclusion Through ‘Alternative Data’. New York: Mastercard
Advisors Research and Analysis.
Martin R (2002) The Financialization of Daily Life. Philadelphia, PA: Temple University Press.
Martin R (2006) Where did the future go? Logos Journal 5(1).
Martin R, Rafferty M and Bryan D (2008) Financialization, risk and labour. Competition and Change
12(2): 120–132.
Mazer R, Carta J and Kaffenberge M (2014) Informed Consent: How Do We Make It Work for Mobile
Credit Scoring? Washington, DC: Consultative Group to Assist the Poor.
Meade K (2015) Credit scoring for inclusive growth. Inclusion Hub, 13 October.
Meade K (2016) EFL Application, Demystified. Cambridge: Entrepreneurial Finance Lab.
Mitchell T (2002) Rule of Experts: Egypt, Techno-Politics, Modernity. Berkeley: University of
California Press.
Mitchell T (2008) Rethinking economy. Geoforum 39: 1116–1121.
Miller P (2001) Governing by numbers: Why calculative practices matter. Social Research 68(2):
Montgomerie J (2008) Bridging the critical divide: Global finance, financialisation, and contemporary
capitalism. Contemporary Politics 14(3): 233–252.
Montgomerie J and Williams K (2009) Financialised capitalism: After the crisis and beyond neoliberal-
ism. Competition and Change 13(2): 99–107.
Multilateral Investment Bank (2015) Psychometrics: Putting Creditworthiness to the Test. Washington,
DC: Inter-American Development Bank.
Muniesa F, Doganova L, Ortiz H, et al. (2017) Capitalization: A Cultural Guide. Paris: Presses de
Nafus D (2014) Stuck data, dead data, and disloyal data: The stops and starts in making numbers into
social practices. Distinktion 15(2): 208–222.
Nafus D and Sherman J (2014) This one does not go up to 11: The quantified self movement as an
alternative big data practice. International Journal of Communication 8: 1784–1794.
National Consumer Law Center (2016) Past Imperfect: How Credit Scores and Other Analytics ‘Bake
In’ and Perpetuate Past Discrimination. Boston, MA: National Consumer Law Center.
Nolke A, Heires M and Bifling H (2013) Editorial: The politics of financialization. Competition and
Change 17(3): 209–218.
Pani R (2014) A fetish and fiction of finance: Unraveling the subprime crisis. Economic Geography
90(2): 213–235.
Pathak P (2014) Ethnopolitics and the financial crisis. The Sociological Review 62: 90–116.
Pellandini-Simanyi L, Hammer F and Varha Z (2015) The financialization of everyday life or the
domestication of finance? Cultural Studies 29(5–6): 733–759.
Philippon T (2008) The Evolution of the US Financial Industry from 1860 to 2007: Theory and Evidence.
New York.
Newswire PR (2015) FICO, LexisNexis Risk Solutions and Equifac Jointing to Generate Trusted
Alternative Data Scores. New York: PR Newswire.
Aitken 299
Samlin S (2016) FinTech Lending Update: Credit Score Alternatives are the Future? Atlanta, GA:
Alston and Bird.
Sassen S (2014) Expulsions: Brutality and Complexity in the Global Political Economy. Cambridge, MA:
Harvard University Press.
Saunders FS (2016) Where on Earth are you? London Review of Books 38(5): 7–12.
Schenker J (2015) Fintech start-ups need not be a threat. Informilo Magazine, 12 October.
Schneider R and Schutte A (2015) The Predictive Value of Alternative Credit Scores. Chicago, IL:
Center for Financial Services Innovation.
Scott J (1998) Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed.
New Haven, CT: Yale University Press.
Steyerl H (2016) A sea of data: Apophenia and pattern (Mis-)Recognition. E-Flux Journal.
Tewari A and Shellenberger D (2014) Alternative Credit Data and FICO Score Fundamentals.
New York: Fair Isaac Corporation.
TransUnion (2016) The State of Alternative Data. New York: TransUnion/Versta Research.
Turner M, Lee A, Schnare A, et al. (2006) Give Credit Where Credit Is Due: Increasing Access to
Affordable Mainstream Credit Using Alternative Data. Washington, DC: Political and Economic
Research Council/The Brookings Institution.
Turner M, Varghese R and Walker P (2015) Research Consensus Confirms Benefits of Alternative Data.
Durham, NC: Political and Economic Research Council.
Van der Zwan N (2014) Making sense of financialization. Socio-economic Review 12: 99–129.
VantageScore (2013) VantageScore 3.0: Better Predictive Ability Among Sought-After Borrowers.
New York: VantageScore.
VantageScore (2015) Maximizing the Credit Universe. New York: VantageScore.
Van Treeck T (2009) The political economy debate on ‘financialization’ – a macroeconomic perspec-
tive. Review of International Political Economy 16(3): 907–944.
VisualDNA (2016) The Big 5. London: VisualDNA.
Walters W (2002) The power of inscription. Millennium: Journal of International Studies 31(1): 83–108.
Walters W (2010) Imagined migration world: The EU and the discourse of anti-illegal immigration.
In: Geiger M and Pecoud A (eds) The Politics of Migration Management. Houndmills: Palgrave,
Wark M (2017) After capitalism, the derivative. Public Seminar, 9 April 2017.
Watkins J (2017) Financialization and society’s protective response: Reconsidering Karl Polanyi’s
double movement. Journal of Economic Issues LI(1): 98–117.
White R (2017) Can the flaws in credit scoring be fixed? The Atlantic, 10 January.
Williams M (2015) FICO testing new ways to assess credit risk. Columbus Dispatch, 1 November.
Wolkowitz E and Parker S (2015) Big Data, Big Potential: Harnessing Data Technology for the
Underserved Market. Chicago, IL: Center for Financial Services Innovation.
Wu C (2015) Proceed with caution on credit scoring with alternative data. American Banker, 11 June.
Yu PM and McLaughlin J (2014) Big Data: A Big Disappointment for Scoring Consumer Credit Risk.
Boston, MA: National Consumer Law Center.
Zaman L and Hawkins K (2015) Using academic factors to reach the invisible borrowers: Could it run
afoul of ECOA? Business Law Today, 21 September 2015, 1–4.
300 Competition & Change 21(4)
... These alternative credit scoring experiments bring greater risk to those who have been made newly 'credit visible' (Aitken, 2017). For example, the opacity in measuring creditworthiness may be risky for young consumers who have no credit data to be captured by the risk assessment system. ...
... The BNPL enterprise represents a particular type of financialization that abandons traditional credit scoring methods in favor of experiments that situate creditworthiness as a dynamic and calculable object. Doing so makes the 'credit invisible' visible using calculative infrastructures that track and gather different data streams that are funneled into opaque, automated credit scoring systems (Aitken, 2017). ...
Full-text available
The rising popularity of fintech‐driven solutions has reshaped financial markets and retail consumer behavior. This paper examines the growing “buy‐now‐pay‐later" (BNPL) phenomenon as a novel, platform‐driven financial innovation to understand how this digitally mediated economic arrangement has produced new financial subjects and subjectivities. Using the idea of 'platform ecologies' that combines the relational focus of financial ecologies with the logics of platform finance, this study highlights how the intermediary role of BNPL services has reshaped relational monetary practices through algorithmically driven modes of operating. Analyzing Singapore’s nascent BNPL landscape through content analysis of BNPL firm websites and critical media coverage of BNPL, this paper shows how automated technologies of risk assessment and debt collection are deployed to govern borrowers and keep them digitally attached to their debts, where creditworthiness is continuously evaluated in response to user transaction and repayment data. The strategic use of affect in framing BNPL services to satisfy immediate materialist consumption masks the fundamental nature of BNPL as debt, while the targeting of young individuals with no credit history through opaque techniques of credit and risk management creates new indebted subjects. The implications of such data‐driven practices in producing greater indebtedness and further shaping financial subjectivities are discussed.
... The data used for statistical credit scoring comprises up to 400 variables provided by credit reference agencies [74]. In the era of big data, often no differentiation between credit data and other data is made and many other variables which are not connected to the credit history of a customer are included [27,[77][78][79][80]. For instance, an applicant's college, her use of capital letters in applications (whereby the use of all caps writing is interestingly a warning sign) and social media data [26], including online tracking and behavioural profiling [79,81]. ...
... Furthermore, other network-based data is included, developing a social credit score based on the individuals' position in a social structure [79]. Additionally, the inclusion of network data allows targeted advertising of credit products to new customers [83] and the inclusion of individuals who previously did not have access to credit [77]. ...
Full-text available
Algorithmic decision-making (ADM) systems increasingly take on crucial roles in our technology-driven society, making decisions, for instance, concerning employment, education, finances, and public services. This article aims to identify peoples’ attitudes towards ADM systems and ensuing behaviours when dealing with ADM systems as identified in the literature and in relation to credit scoring. After briefly discussing main characteristics and types of ADM systems, we first consider trust, automation bias, automation complacency and algorithmic aversion as attitudes towards ADM systems. These factors result in various behaviours by users, operators, and managers. Second, we consider how these factors could affect attitudes towards and use of ADM systems within the context of credit scoring. Third, we describe some possible strategies to reduce aversion, bias, and complacency, and consider several ways in which trust could be increased in the context of credit scoring. Importantly, although many advantages in applying ADM systems to complex choice problems can be identified, using ADM systems should be approached with care – e.g., the models ADM systems are based on are sometimes flawed, the data they gather to support or make decisions are easily biased, and the motives for their use unreflected upon or unethical.
... Massive databases are aggregated and explored for identifying patterns and correlations that were previously impossible (boyd and Crawford, 2012;Beer, 2018). For instance, credit reporting agencies have expanded scoring systems by gathering data relating to non-financial payments, digital footprints, and psychometric testing (Aitken, 2017). Media industries also adopt digital tools to improve audience analytics (Livingstone, 2019). ...
Full-text available
This article examines citizen scoring in China's Social Credit Systems (SCSs). Focusing on 50 municipal cases that potentially cover 210 million population, we analyze how state actors quantify social and economic life into measurable and comparable metrics and discuss the implications of SCSs through the lens of social quantification. Our results illustrate that the SCSs are envisioned and designed as social quantification practices including two facets: a normative apparatus encouraging “good” citizens and social morality, and a regulative apparatus disciplining “deviant” behaviors and enforcing social management. We argue that the SCSs illustrate the significant shift in which state actors increasingly become data processors whereas citizens are reconfigured as datafied subjects that can be measured, compared, and governed. We suggest that the SCSs function as infrastructures of social quantification for enforcing social management, constructing differences, and nudging people towards desired behaviors defined by the state.
... The critical data studies literature often describes datafication in the Global South as a form of capitalist extraction or 'surveillance capitalism', and sometimes termed it 'data colonialism' (Aitken, 2017;Sadowski, 2019;Thatcher et al., 2016;Zuboff, 2015). It argues that, although legal in the strict sense of the term, it is still highly problematic how global corporations acquire data as a commodity in unfavourable contractual transactions either in exchange for using a service or by explicitly paying for it (Elvy, 2017(Elvy, : 1407, relying on end-user license agreements (EULAs). ...
Full-text available
The collection, processing, storage and circulation of data are fundamental element of contemporary societies. While the positivistic literature on ‘data revolution’ finds it essential for improving development delivery, critical data studies stress the threats of datafication. In this article, we demonstrate that datafication has been happening continuously through history, driven by political and economic pressures. We use historical examples to show how resource and personal data were extracted, accumulated and commodified by colonial empires, national governments and trade organizations, and argue that similar extractive processes are a present-day threat in the Global South. We argue that the decoupling of earlier and current datafication processes obscures the underlying, complex power dynamics of datafication. Our historical perspective shows how, once aggregated, data may become imperishable and can be appropriated for problematic purposes in the long run by both public and private entities. Using historical case studies, we challenge the current regulatory approaches that view data as a commodity and frame it instead as a mobile, non-perishable, yet ideally inalienable right of people.
The recent pandemic saw the operations of many businesses shifting to virtual mode. Tasks like psychometric analysis of individuals, for various applications, are conducted online. In this article, we introduce a novel system to analyze the semantics of an individual's tweets from their Twitter profile using LIWC and SALLEE scores. These scores can be used to evaluate less fortunate, thin‐filed candidates using their Twitter profiles. With increased access to phones and the internet, many organizations are focusing on making credit systems available to the masses by introducing psychometric analysis. This article proposes a dynamic model for evaluating the personality of a Twitter user using the textual content shared on their page. The model will allow stakeholders to ascertain the personality of user according to any personality model. To analyze if this is viable and flexible approach to model any kind of personality model, we take MBTI personality dataset and train classifier to predict personality types. Then these results are correlated with a linguistic score to find correlation between the two. We found that proposed approach, outperformed the other relevant works also some aspects of these linguistic scores show a heavy correlation with certain personality types.
Kenya is a widely cited case for proponents of fintech for development. This article shows how Kenya’s fintech boom replicates patterns of uneven development inherited from the colonial era. In particular, fintech use is unevenly distributed between urban and rural areas, and heavily concentrated on Nairobi and Mombasa in particular. The article seeks to explain these patterns by situating them in relation to the spatiality and political economy of settler‐colonial agriculture, tracing successive (unsuccessful) efforts at reforming the financial system to ameliorate social and spatial disparities inherited from the colonial era. It does so drawing on recent debates about “financial infrastructures”, alongside considerations of the political economy of land, property relations, and the state.
Full-text available
This article traces experiments aimed at promoting wider adoption of ‘microinsurance’ – small, simplified insurance policies targeting the poorest. Microinsurance is a central element of a wider turn towards the promotion of ‘resilience’ in global development. The development of commercial markets for microinsurance, however, has failed to meet the expectations of promoters. This article traces the ways that the diverse donor agencies, professional organizations and philanthropic organizations involved in the promotion of microinsurance have responded to these failures, primarily by seeking to articulate basic data infrastructures that might make possible profitable insurance operations. These activities are described as a kind of ‘anticipatory marketization’ – experiments seeking to prepare the ground for the emergence of markets for risk management, thus far without much success. Where microinsurance has often been described in terms of ‘financialization’, this article suggests that there are important political dynamics at play that have been overlooked. Efforts to develop markets for microinsurance, and the persistent focus on troubleshooting and re-engineering those markets in the face of failure, are not driven directly by finance capital. Rather, they reflect fraught efforts to articulate modes of social protection not requiring substantial redistribution.
Social media commentary, satellite imagery and GPS data are a part of ‘alternative data’, that is, data that originate outside of the standard repertoire of market data but are considered useful for predicting stock prices, detecting different risk exposures and discovering new price movement indicators. With the availability of sophisticated machine-learning analytics tools, alternative data are gaining traction within the investment management and algorithmic trading industries. Drawing on interviews with people working in investment management and algorithmic trading firms utilizing alternative data, as well as firms providing and sourcing such data, we emphasize social media-based sentiment analytics as one manifestation of how alternative data are deployed for stock price prediction purposes. This demonstrates both how sentiment analytics are developed and subsequently utilized by investment management firms. We argue that ‘alternative data’ are an open-ended placeholder for every data source potentially relevant for investment management purposes and harnessing these disparate data sources requires certain standardization efforts by different market participants. Besides showing how market participants understand and use alternative data, we demonstrate that alternative data often undergo processes of (a) prospecting (i.e. rendering such data amenable to processing with the aid of analytics tools) and (b) assetization (i.e. the transformation of data into tradable assets). We further contend that the widespread embracement of alternative data in investment management and trading encourages a financialization process at the data level which raises new governance issues.
Full-text available
Hundreds of millions of people in low-income economies do not have a credit or bank account because they have insufficient credit history for a credit score to be ascribed to them. In this paper, we evaluate the predictive accuracy of models using alternative data, that may be used instead of credit history, to predict the credit risk of a new account. Without alternative data, the type of data that is typically available is demographic data. We show that a model that contains email usage and psychometric variables, as well as demographic variables, can give greater predictive accuracy than a model that uses demographic data only and that the predictive accuracy is sufficiently high for the demographic and email data to be used when conventional credit history data is unavailable. The same applies if merely psychometric data is included together with demographic data. However, we show that different randomly selected training: test sample splits give a wide range of predictive accuracies. In the second part of the paper, using two datasets that include only email usage as a predictor, we compare the predictive performances of a wide range of machine learning and statistical classifiers. We find that some classifiers applied to these alternative predictors give sufficiently accurate predictions for these variables to be used when no other data is available.
Full-text available
What is social visibility? How does it affect people and public issues? How are visibility regimes created, organized and contested? Tackling both social theory and social research, the book is an exploration into how intervisibilities produce crucial sociotechnical and biopolitical effects.
Financialization challenges Karl Polanyi’s thesis of double movement, the thesis that efforts to extend the market evoke efforts to protect humans, nature, and means of production from market forces. Financialization refers to the increased power of financial institutions. The government protects the incomes and assets of financial institutions, but it does little to protect the incomes and assets of households, which are necessary for people to afford healthcare, education, emergencies, retirement, and so on. Polanyi criticized nineteenth-century civilization for transforming land, labor, and the means of production into commodities, using economic insecurity to motivate humans. The development of intangible property allowed business to expand the market in two ways: (i) restricting output to drive up profits and (ii) liquefying consumer assets to provide credit to consumers to increase spending. The implications of that process manifested themselves in the financial crisis of 2008. Market capitalism represented the attempt to organize commodities based on economic rationality. Similarly, the twentieth- and twenty-first-century capitalism represents the effort to “rationally” organize society according to the value of intangible assets. Both efforts failed, indicating the continued relevance of Polanyi’s thesis.
During the last 30 years, finance has increased not only its share of economic activity but also of people's aspirations. This has transformed society by increasingly organizing it around the search for financial efficiency. Is a society based on fundamental values of free judgment, responsibility and solidarity still possible?.