A Discussion Paper on the Grey Area The Ethical Problems Related
to Big Data Credit Reporting
Victor Chang and Jing Lin
International Buiness School Suzhou, Xi’an Jiaotong-Liverpool Universiy, Suzhou, China
Keywords: Big Data Credit Reporting, Ethical Concerns for Big Data, Controversies on Big Data Uses.
Abstract: With the rise and the development of the “credit society”, the credit reporting has played a central role in
evaluation one’s credit statues, including monitoring and updating creditworthiness of individuals. As the
emergence of big data, new tools enabling the credit reporting system to develop new level, by collecting the
online and offline data to establish more completely score system. This review paper is aimed to present the
difference between the new big data credit reporting and traditional credit reporting, and then explain
advantages offered by the new data management. Subsequently, ethical problems will be described due to
rising concerns. Being “kidnapped” by the credit reporting applications, users’ data will be collected and
disposed without prior permission. Some data processes may arise with the messy, unreasonable and fake data
resource problems to add more complexities to the existing services which are unable to cope with. As a
result, individual users could not verify the correctness of the data and did not know which data would be
more trustworthy to be verified for payment and billing. To be worse, users even do not know how to improve
their creditworthiness if they have done everything correctly. There are some issues about precision
marketing, since some data brokers will target the individuals who was vulnerable to the non-performing and
short-term loans. Last but not least, the algorithm of big data prone to evaluate the credit score by groups that
individual related to, rather than the individual’s own merits, which may lead to discrimination issue, and
accelerate the wealth gap problem.
1 INTRODUCTION
Every citizen in USA have a unique society security
number, which recorded all the credit activities,
include pay off the credit card debt, apply for the
mortgage or student loans (Komuves, 1997). There
are three main credit bureaus in the American,
Equifax, Experian and TransUnion, they are all
owned and operated by publicly companies, not
owned by any governments. According to your credit
history, amounts owed and debts on records, they will
generate an individual credit reporting, it can be
corresponded to your society security number, which
means once you were recorded, it would not
disappear. With this credit reporting, you can apply
for bank loans or jobs. If you were identified as an
individual with a poor credit record, the bank will
reject to give you loans and the employer will reject
you either. Before the wide adoption of big data
technology, the credit reporting is mainly used in the
financial field to evaluate applicants’ ability to repay
their loans (Mierzwinski and Chester, 2013), with
more social data and information, the credit reporting
system grew into the creditworthiness evaluation
system, which can be used in all aspects of life
(Miller, 2000). One of the big data reporting strengths
is the huge amount data resources that it can get
connected, queried and linked altogether. Another
strength is the machine learning and AI technologies
to support large scale data processing, analysis,
categorization and management. Unlike the
traditional data resource, the ‘thin-file’ customers
who have less credit history due to age, immigrant
status or recent credit record lacking, will get a fair
credit score in big data credit reporting (Bureau,
2012). Using the big data services, which can
combine as much as relevant credit data, such as the
social data, property and online activities. By cross
checking to find whether there are no bad records, the
thin-file customers can get loans by a lower rate.
However, as an idiom has rightly pointed out - the
water that bears the boat can be the same that
swallows it up. The rise of quantity, variety and
complexity in data can lead to chaos, if it is not
348
Chang, V. and Lin, J.
A Discussion Paper on the Grey Area The Ethical Problems Related to Big Data Credit Reporting.
DOI: 10.5220/0006823603480354
In Proceedings of the 3rd International Conference on Internet of Things, Big Data and Security (IoTBDS 2018), pages 348-354
ISBN: 978-989-758-296-7
Copyright
c
2019 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
properly managed. In the data collection and
transformation, most people will face the all-or-
nothing decision. The user will not have access to
services if he/she rejects to upload the private
information, such as locations, or delivery addresses.
Although some application suppliers announce that
the application will not use the customers’ data for
profits, their authenticity is debatable. There are
already some cases of erroneous data problems, two
women’s credit reporting have been swapped because
they have similar personal information. In addition,
some researchers founded that there exist the
fabricated issues in the data input procedures. While
putting the subjective views aside, data problem may
be structured because of a certain level of error rate in
model operations (Pasquale, 2015). The Federal
Trade Commission have showed in the report that the
error rates up to 26%. The structural problem also
related to the machine learning processes, due to the
complex algorithms, the machine will decide how
much attention an emphasis it needs to give and
dynamically adjust the system itself regularly. It
would be difficult to perform manual regulations due
to millions or billions of users involved.
Additionally, machine learning and AI techniques
may not be robust enough to provide secure services.
Some developers can use the tools to aim at the
vulnerable, high-value targets for non-performing
loans, just like sell heroin to drug addicts
(Mierzwinski and Chester, 2013). Furthermore, the
machine learning can also assess the individual’s
credit situations by evaluating their circles of friends,
groups and communities of politics, religions, and
others. Instead of evaluating the individual’s own
merits, it is unfair to the vulnerable groups. As the
situation continues to deterioration, the wealth gap
will be widened, for the wealthy people will get more
service and better remedial measure to keep their
credit reporting more dignity.
2 LITERATURE REVIEW FOR
BIG DATA CREDIT
REPORTING
Hilda et al. (2016) indicated the benefit of big data
used in the credit reporting field, in his review, he
illustrated some new data processing methods, which
can be profitable in business. Singer (2012) explain
some new data source, such as the web browsing data
or user purchase records, can be used in the credit
evaluation process. The new data added in can help
the thin-file people get reasonable evaluation
(Bureau, 2012). While in the data collection, user will
be forced to make an all-or-nothing choice between
the acceptance of free services or refuse to be
recorded (Bilogrevic et al., 2014). The Federal Trade
Commission (2013) also points out that credit score
counting process with machine learning contain a
certain percentage of errors. Jacobs (2015) manifests
that the online data has inaccuracy problems. Hurley
and Adebayo (2016) claims that individuals could not
improve their credit score due to the non-transparent
issue. Furthermore, some “precision marketing” will
target the people who were vulnerable to the high
interest rates but short-term loans, which will indulge
the high-risk borrowers (Trusts, 2014). As the
machine will be prone to assess the individual’s
credits by the group-referential processing, the
discrimination issue will be deteriorated (Meyer,
2015). When such vicious cycles repeat more often,
wealth gap will be widened (Hurley and Adebayo,
2016).
Before 1980s, lending a loan is decided by
individual loan officers and specialists who assess
applicants, which can differ from person to person
(Citron and Pasquale, 2014). Then, the decision for
lending based on the automated credit scoring
systems developed by the Fair and Isaac
Corporation(FICO). Moreover, the credit scoring is
more useful, it uses sum of a person's apparent
creditworthiness to make underwriting decisions,
similar to "predict the relative likelihood of a negative
financial event, such as a default on a credit
obligation." (Yu, 2014). In America today, every
citizen has equipped with one unique social security
number, which record all the credit activities, such as
paying off credit card and loan, repaying the
education loans. Customers can get their credit
reporting from three main credit bureaus in America,
Equifax, Experian and TransUnion, they both
publicly-traded, not owned by government. For the
most of Americans, without a good credit score will
cause many inconveniences, such as fail to get a
mortgage, fired by employer, or even lose the
opportunity for education. As a result, it is essentially
important to pay attention to individual credit score.
Nowadays, Hilda et al. (2016) claimed that use of
big data has been adopted by more organizations.
This can be done by identifying patterns, clusters and
outliers that are not obvious by the traditional
methods. The ultimate goal is to bring profits to
business. As the big data technology has becomes
more mature, it can handle the massive amounts data
with high speed. The field of credit reporting has been
reforming.
A Discussion Paper on the Grey Area The Ethical Problems Related to Big Data Credit Reporting
349
2.1 The Big Data Credit Reporting
Different from Traditional One
2.1.1 Data Source
The big data credit reporting is different from the
traditional one, the first part is data source, the big
data credit reporting can utilize the different kinds of
non-traditional data, which include the criminal
records, social media data, consumers’ retail
spending history, the online purchase records,
internet browsing history, or even an individual's
friend circle on social networking board (Singer,
2012). Although the traditional data sources still have
the basic of credit system, the new alternative tools
equipped with the big data has rapidly occupying the
market. Like the Experian, it has used big data to
exploit thousands of ‘universal customer profiles’
that can integrate both the online and offline data
(Tewksbury, 2013). As same as the Fair and Isaac
Corporation(FICO) the main planner of lending
decisions has exploit a new system combined with
non-traditional data to assess the thin-file borrowers,
for they lack the traditional data and difficulty to
evaluate. The new ‘FICO Score XD’ is the result of
the cooperation between the FICO and credit bureau
Equifax, they use the cable and mobile phone
accounts to predict the consumers’ creditworthiness
(Carrns, 2015). Nowadays, the most remarkable
credit scoring and underwriting company is the
ZestFinance, who helps two groups of people getting
loans. One group includes the people who cannot
meet the basic requirements of lending because their
FICO score is less than the 500, and other group who
have high loan cost and low credit score. It uses the
proprietary algorithms to analyse several thousand
data points per individual in order to arrive at a final
score (Lohr, 2015).
2.1.2 Machine Learning
The new method adopted by ZestFinance was based
on the machine learning, which is also the different
part with the traditional credit reporting system.
Machine learning is a method which can
automatically explore the data patterns and used the
pattern to predict the future data (Robert, 2012).
There are two styles of machine learning, supervised
and unsupervised. The supervised one try to find out
the relationship pattern between the target which the
data analyst already confirmed and the other data
points collected. Unsupervised try to predict the
target variable, assisting analyst understand how the
data was generated and discovered the pattern behind
them (Calders and Custers, 2013). The big data credit
reporting belongs to the use of supervised machine
learning tools.
2.2 The Big Data Credit Reporting’s
Characteristic and Advantage
With numerous data sources and machine learning
tool, the big data credit reporting has been rapidly
developed. One reason for traditional credit reporting
been replaced is that the big data credit reporting can
predict the thin-file consumers, the people who
cannot get access to the credit score, such as many
immigrants or recent college graduates with little or
no credit history, or people who have not activated its
credit account at least six months. With the big data
tool, equipped with huge data and high speed, these
people can also get the credit derives from their
normal life activities. Another reason is its foresight,
it combines as much as possible relative credit data,
such as the usage time of the same phone number,
connections on Facebook, a stable address or even the
use of proper capitalization in filling out a form, they
can assess the credit in every aspect. The last but not
the least reason is that it can protect the vulnerable
groups, for example, LexisNexis had created a new
credit score system called RiskView (Bureau, 2012),
this product include the traditional public record like
the foreclosures and bankruptcies, in addition, it also
include the educational history, professional licensure
data, and personal property ownership data, as a
result, these people who don’t have the traditional
credit score but have an professional license or pay
rent on time, or own a car, may get a better access to
credit system than they otherwise would have
(Ramirez, 2016). The Future of Privacy Forum
(Polonetsky, 2014) report also indicated that business
and government use the big data to protect and
empower the disadvantaged group, giving them the
access to job markets, disclosure the discriminatory
practices and improve the education and help those in
need.
3 THE ETHICAL PROBLEM
ABOUT THE BIG DATA
CREDIT REPORTING
3.1 The Big Data Credit Reporting
Different from Traditional One
When you use mapping software it might record all
your footmarks and sell it without your permission, if
SPBDIoT 2018 - Special Session on Recent Advances on Security, Privacy, Big Data and Internet of Things
350
you don’t give the application permission to access
your information, then you cannot use the function,
you cannot control what information would be
collected and often be forced to make all-or-nothing
decision between receiving free services or refusing
to be profiled (Bilogrevic et al., 2014). Not only a
fraction of people has such thoughts, the
investigations revealed that, seven in ten Europeans
are worried about that software operator will use the
personal data for profit or other uncontrolled use
(Social, 2011). Over the past few years, the online
applications have collecting the user’s information to
build the individual profiles, then selling them to the
advertiser and data broker to make profit, the users
have little control with the data collect process.
3.2 The Messy, Unreasonable and
Faked Data
Some aggressive big data enthusiasts claimed that we
should embrace the ‘messy data’, for the errors in data
manipulated process can help develop better pattern
result, the more mistakes there are, the more
preparations and well-equipped the system has gone
through (Alloway, 2015). However, the messy
problem had already affected the normal person life.
There is one case, the victim is Judy Thomas, who
sued Trans Union for regularly mixing her report with
Judith Upton. One day in 1996, Judy found there were
some mistakes since some bad debts appeared on her
credit report. After the investigation, she found the
bad debts was belonging to a woman named Judith
Upton, the one who has similar first name and same
birth year with her, and only one number different in
their Social Security numbers. Due to the operation of
one careless staff, Judy could not get the loan based
on her expectancy value, she struggled to appeal for
justice. Finally, Judy was awarded more than 5
million dollars by a Federal Court jury as a
compensation (Weisman, 2013). In this case, it
reveals problems with the quality of data in credit
reporting system as follows, first, erroneous data
could be caused by some careless problems, second,
there were fake and fabricated data, researchers found
that some investigators of the credit system even
fabricated derogatory information about individuals.
Apart from quality of data, the lack of third party
central regulation on credit agencies appear to be an
issue. While the credit rating agencies do not have the
direct relations with the customs, they lack incentives
to treat the individuals fairly, if profits and finding
vulnerable individuals appear to be their concerns.
There is also another challenge - it is difficult to
interchange with and between the credit rating
agencies. To sum up all the observations above, the
data problem is not just an anecdotal, it is structural
(Cetorelli and Gambera, 2001; Kenny, 2014). In a
detailed and comprehensive investigation conducted
by scholars and Federal Trade Commission, there are
almost 3,000 credit reports belonging to 1,000
consumers, it has been found that 26 percent had
“material” error problems which were serious enough
to affect the individuals credit scores. Even with the
conservative estimates, there were still 23 million
Americans having errors in their credit reports. Once
their credit scores amended, they will obtain credit
loan in lower prices (commission, 2013). On the other
hand, some researchers and industries might think the
big data can improve the situations, since more
information can be cross checked and validated for
better precision and accuracy. Some data may seem
logically related with the credit record, for example,
payment historical records matches the identity of the
right person. However, there are other “fringe data”
like the reading time of user notice to indicate the
individual’s degree of care might lack the correlations
with the creditworthiness (Yu, 2014). Additionally,
lots of evidence confirm that since thousands of data
were from the individuals on and off the site
activities, it might result in getting a high percentage
of inaccurate information. As Jacobs (2015) claimed
that mobile location data can be easily lead to
inaccuracy. Inaccurate data problems can deeply
affect the individual credit activities. While the big
data can improve on speed and interconnectivity, it
might less efficient to double check accuracy and
validity of the data, the problem may be worse.
3.3 Non-Transparency
The “kidnap service” and data source problem both
brought out one issue - individuals cannot get access
to the data process. The operation processes are
opaque to customers, like the case about Kevin
Johnson, who was a person with a good and decent
credit. He received a letter from the American
Express in one day of late 2008, which told him that
the credit limit of his credit card had been decreased
from $10800 to $3800. The reason given by
American Express was that, the market which Kevin
recently shopped by have some customers who have
bad credit history with American Express. While
Kevin casted around for an explanation, the American
Express did not want to share more details about that.
Even the Federal Trade Commission have sued the
three major credit rating agencies in 2000, because
they did not answer phones, which means the
customers have always been neglected. If users did
A Discussion Paper on the Grey Area The Ethical Problems Related to Big Data Credit Reporting
351
not know how their scores been calculated and even
cannot complain about the unfair scores, the fairness
of credit reporting can be questionable. On the basis
of electronic privacy information centre, big data
credit evaluators are mainly "concerned about
collecting a large amount of information about
individuals" and the overall quality of the data may
be affected (Scoring, 2016). With non-transparent
problems exist, individuals cannot identify the unfair
credit scores and fail to improve scores or prevent
them from falling further (Hurley and Adebayo,
2016). The non-transparent problems exist in two
aspects, one is in the data collection and
transformation process, another is due to the machine
learning algorithms. Guzelian et al. (2015) explain
that the credit companies treat their data as the
commercial secrets. Therefore, the data collection
progress would be opaque, in case the competitor
found out data resources, the customers could not get
access to the data, they could not find whether the
data was accurate either. Even if the customers know
the data is accurate, it is difficult to find one error in
the millions of entries in the credit scorer is raw data
set. Since the machine learning process is designed to
find the relationship between data and target, the data
collected from customers will be transferred to the
language that computers can understand, the process
will be involved with mass aggregations and
combinations of data points. In this case, if the
machine learning algorithm was a “layman”, then the
layman could not understand how much a dataset
would need and what types of diagnosed emphasis for
further actions. As a result, it is not straight forward
to get some effective evidence to identify the
algorithm validity.
3.4 Heroin for Drug Addicts
Trusts (2014) shows the lenders will use the most
advanced technology and complex algorithm to target
the vulnerable persons, particularly those attracted by
the low-valued, short-term credit products with
usurious interest rates and highly adverse terms. The
experts at Upturn have also supported this view.
There is potentially one possible situation: While
some credit reporting system developers will not
interest in the system improvement or predicting the
consumer creditworthiness, they choose to put efforts
on finding vulnerable or high-value targets for non-
performing loans. The survey shows the target
disproportionately come from poor and minority
communities (Hurley and Adebayo, 2016). Angwin
(2014) identified that although no unambiguous
evidence showed the big data is used in the identified
the vulnerable borrower, some major data brokers
who work on credit reporting system have been
accused for bringing the "sucker lists" to the market,
the list which specifically target on vulnerable groups
of people, which may cause them distress or more
pain to their existing debts. The evidence can be
found from 2013 Senate Commerce Committee
report which have lists with title like "Hard Times",
"Burdened by Debt", "Retiring on Empty" and "X-tra
Needy" which were deliberately aimed at the
individuals who were trying to buy some
unfavourable financial products.
3.5 Discrimination
Facebook is another good example. It is a free social
networking website which can allow users to upload
their photos or videos, and post their comments, has
recently applied for one patent application. The patent
is related to one method for "authorization and
authentication based on an individual's social
network." which means the users’ social information
can be used to evaluate their credit score. The patent
application explains that, when someone applies for
loan, the lender can get access to the individual’s
social network profile and get the rank of his social
group, if the social group meet the minimum
requirement of the lender, then the lender can give a
loan to the applicant, otherwise, he will be rejected.
Criticasters have indicated that the tools can bring
about the new style of digital redlining (Meyer,
2015).
The machine learning may also produce results
that have inequity biases, because the person’s final
credit score is not based on the individual’s own value
but prefer on the basis of the relative group that the
system affirm effectively. As Barocas and Selbst
(2016) claim that when a model relies on the
generalization reflected in the data, the final result of
individual maybe statistically sound inference, so that
the result may be inaccurate. However, this may
happen when customers do nothing with it and they
cannot do anything to revert. Like the case of Kevin
Johnson mentioned earlier, the model will punish
individuals that in a particular environment like the
designated community or having some issues with the
political or religious grounds. The case of Kevin
Johnson is not a special case, in lots of other fields,
from the school admission decisions to the insurance
selling, the big data tool will judge the person by the
shared data rather than individuals' own and true
value, which have been proven to aggravate existing
prejudice (Lowry and Macpherson, 1988).
SPBDIoT 2018 - Special Session on Recent Advances on Security, Privacy, Big Data and Internet of Things
352
3.6 Wealth Gap Widen
Schmitz (2014) claimed that the use of big data in the
credit reporting field can foster the discrimination,
due to the data analysis subjective rules the machine
will give an aid to the smartest customers with the
best services and best remedial measure, which will
widen the gap between the wealthy and poor
individuals. As the Hurley and Adebayo (2016)
explained in the report, the big data tools will increase
the risk of creating a system of "creditworthiness by
association", which combine the individuals
families, religions, sexual orientation and other
relevant information. All information will figure out
the individual’s credit score, whether you can pay for
a load is not determined by how much you can earn,
but the group you are in, which will further aggravate
discrimination. Like the example of Kevin Johnson,
he became the proxies of sensitive information like
the race and vulnerable attributes. In this example, the
big data could not eliminate the prejudice, but
aggravate the existed bias.
4 CONCLUSION
In this paper, we report the grey area about the use of
big data, particularly for credit reporting system that
has biases and selection on vulnerable groups of
people. Big data enthusiasts argue that all the data can
be managed by credit reporting and the big data will
lead us to live happily and comfortably. While we can
find is that the data maybe objective, the process of
data categorization and evaluation can be subjective.
Moreover, the ways data will be collected or
diagnosed largely depend on the existing social
resources. Individuals who have contributed their
data to the credit reporting system, have no or few
effective actions to modify the “unreasonable” data.
The vulnerable people will be likely to be trapped in
a vicious circle by precision marketing or have been
considered as low credit groups due to potential
discrimination in income, social status, race, politics
or religion. On the other hand, the wealthy people can
possess the best remedial measures and excellent
services. While we need to use the big data to build
the completeness and soundness of the credit
reporting system. However, we should also pay
attention to ethical issues raised by the big data and
find betters to manage and treat each individual fairly
and equally.
REFERENCES
Alloway, T., 2015. Big data: Credit where credit's due
[Online]. Available from:
https://www.ft.com/content/7933792e-a2e6-11e4-
9c06-00144feab7de (Accessed: 2017).
Angwin, J., 2014. Dragnet nation: A quest for privacy,
security, and freedom in a world of relentless
surveillance. Macmillan.
Barocas, S., Selbst, A. D., 2016. Big data's disparate
impact’, Cal. L. Rev., 104, p. 671.
Bilogrevic, I., Freudiger, J., De Cristofaro, E., & Uzun, E.
(2014) ‘What’s the gist? privacy-preserving
aggregation of user profiles’, In: KUTYŁOWSKI, M.
& VAIDYA, J. (eds.) Computer Security - ESORICS
2014: 19th European Symposium on Research in
Computer Security, Wroclaw, Poland, September 7-11,
2014. Proceedings, Part II. Cham: Springer
International Publishing.
Bureau, C. F. P., 2012. Analysis of Differences between
Consumer-and Creditor-Purchased Credit Scores.
Calders, T. & Custers, B., 2013. What Is Data Mining and
How Does It Work? In: Custers, B., Calders, T.,
Schermer, B. & Zarsky, T. (eds.) Discrimination and
Privacy in the Information Society: Data Mining and
Profiling in Large Databases. Berlin, Heidelberg.
Carrns, A., 2015. New Credit Score Systems Could Open
Lending to More Consumers [Online]. Available from:
http://lippmancpas.com/new-credit-score-systems-
could-open-lending-to-more-consumers (Accessed:
2017).
Cetorelli, N., & Gambera, M., 2001. Banking market
structure, financial dependence and growth:
International evidence from industry data’, The Journal
of Finance, 56(2), pp. 617-648.
Citron, D. K., & Pasquale, F., 2014. ‘The scored society:
due process for automated predictions’, Wash. L. Rev.,
89, 1.
Commission, F. T., 2013. Section 319 of the Fair and
Accurate Credit Transactions Act of 2003: Fifth Interim
Federal Trade Commission Report to Congress
Concerning the Accuracy of Information in Credit
Reports.
Guzelian, C. P., Stein, M. A., & Akiskal, H. S., 2015.
‘Credit Scores, Lending, and Psychosocial Disability’,
BUL Rev., 95, p.1807.
Hilda, J. J., Srimathi, C., & Bonthu, B., 2016. ‘A review on
the development of big data analytics and effective data
visualization techniques in the context of massive and
multidimensional data’, Indian Journal of Science and
Technology, 9(27), pp.1-13.
Hurley, M., & Adebayo, J., 2016. Credit scoring in the era
of big data’, Yale JL & Tech., 18, p.148.
Jacobs, S., 2015.Report: More Than Half of Mobile
Location Data is Inaccurate [Online]. Available from:
http://streetfightmag.com/2015/05/14/report-more-
than-half-of-mobilelocation-data-is -inaccurate
(Accessed: 2017).
Big data: a tool for fighting discrimination and empowering
groups (2014) [Online]. Available from:
A Discussion Paper on the Grey Area The Ethical Problems Related to Big Data Credit Reporting
353
https://fpf.org/2014/09/11/big-data-a-tool-for-fighting-
discrimination-and-empowering-groups (Accessed:
2017).
Kenny, D. A., 2014. Measuring model fit.
Komuves, F. L., 1997. ‘We've Got Your Number: An
Overview of Legislation and Decisions to Control the
Use of Social Security Numbers as Personal
Identifiers’, J. Marshall J. Computer & Info. L., 16,
p.529.
Lohr, S., 2015. Big Data Underwriting for Payday Loans
[Online]. Available from: http://bits.blogs.
nytimes.com/2015/01/19/big-data-underwriting-
forpayday-loans/? r=0 (Accessed: 2017).
Lowry, S., & Macpherson, G., 1988. ‘A blot on the
profession’, British medical journal (Clinical research
ed.), 296(6623), p.657.
Meyer, R., 2015. ‘Could a Bank Deny Your Loan Based on
Your Facebook Friends?’, The Atlantic.
Mierzwinski, E., & Chester, J., 2013. ‘Selling consumers
not lists: the new world of digital decision-making and
the role of the Fair Credit Reporting Act’, Suffolk UL
Rev., 46, p.845.
Miller, M., 2000. ‘Credit reporting systems around the
globe: the state of the art in public and private credit
registries’, World Bank. Presented at the Second
Consumer Credit Reporting World Conference, held in
San Francisco, California, October.
Pasquale, F., 2015. The black box society: The secret
algorithms that control money and information’,
Harvard University Press.
Robert, C., 2014. Machine learning, a probabilistic
perspective.
Ramirez, E., 2014. Big Data: A Tool for Inclusion or
Exclusion? US FTC.
Schmitz, A. J., 2014. ‘Secret Consumer Scores and
Segmentations: Separating Haves from Have-Nots’,
Mich. St. L. Rev., p.1411.
Scoring, C., 2016. Electronic Privacy Information Center
Singer, N., 2012. ‘Mapping, and sharing, the consumer
genome’, New York Times.
Social, T. O., 2011. Special Eurobarometer 359 Attitudes
on Data Protection and Electronic Identity in the
European Union.
Tewksbury, M., 2013.The 2013 Big Data planning guide
for marketers. Experian Marketing Services.
Trusts, P. C., 2014. Fraud and abuse online: Harmful
practices in internet payday lending. Washington, DC.
Weisman, S., 2013. A Guide to Elder Planning: Everything
You Need to Know to Protect Your Loved Ones and
Yourself. FT Press.
Yu, R., 2014. Knowing the Score: New Data, Underwriting,
and Marketing in the Consumer Credit Marketplace.
SPBDIoT 2018 - Special Session on Recent Advances on Security, Privacy, Big Data and Internet of Things
354