Alternative Credit Scoring

Concept, Practice, and Global Developments

Dr Ashish Srivastava
AGM/ Member of Faculty
College of Agricultural Banking,
Reserve Bank of India,
University Road,
Pune – 411016

Email: [email protected]
Phone: +91 (20) 2558 2325

The views are personal and do not necessarily reflect those of the Reserve Bank of India.
 Alternative Credit Scoring
Concept, Practice, and Global Developments

I. INTRODUCTION
The credit risk, defined as the probability of incurring a loss due to failure of a borrower or
counterparty to meet its repayment obligations in accordance with the agreed terms, is the
most significant risk in banking. Credit risk may be involved in a number of ways and has the
potential to transform into other risks, such as market risk and liquidity risk. As such, the
credit risk management encompassing identification, measurement, mitigation and control of
the credit risk remains on the top priority of the bankers. Credit scoring plays a crucial role in
this process. However, the traditional approaches to credit scoring are increasingly being
found inadequate and hence, the consequent quest for alternative approaches and new
perspectives. A large quantum of technical work is underway in the area of alternative credit
scoring to better understand and put in place an effective process of the credit risk
measurement and management. However, the traditional and new approaches have their
complementarities and inter-dependence. Use of the alternative data and application of the
behavioural sciences have added to the scope and utility of the alternative credit analysis and
scoring. This paper provides a broad overview about the subject while presenting an insight
into the latest developments in the domain of alternative credit scoring.

II. CREDIT SCORING

Accurate measurement and assessment of credit risk is critical as the quality of credit
exposure is one of the riskiest areas for banks. Accordingly, credit assessments have to be
sophisticated, directed and streamlined. Over time, as credit risk has become more
diversified and complex, the importance of accurate credit risk assessments and the need to
understand the main components that drive the credit risk has increased. Correct pricing and
management of credit risk exposures hinge on their precise assessment. Similarly, the
supervisory authorities, who have to ensure that the banks' credit risk management systems
are sufficiently safe and sound, precise measurement of credit risk is important as any
deficiency in this regard may jeopardize a bank's profitability and ultimately its survival,
which may also assume systemic proportion. However, it is not very easy to standardize
the

1
circumstances leading to defaults and hence, detailed specific analysis and foresight is needed
instead of having a broad-brush approach.

Credit scoring as a tool of credit appraisal helps to understand and predict the likely
behaviour of the prospective borrowers in respect of the repayment and adherence to the
terms of the agreement of a credit proposal based upon certain parameters and assessments
made thereof. Essentially, the credit analysis captures the information about five C’s of a
potential borrower, namely, character, capacity, capital, collateral, and condition. Credit risk
measurement depends upon the three key components, namely, Probability of Default (PD),
Exposure at Default (EAD) and Loss Given Default (LGD). Credit appraisal help in assigning
a probability for the likelihood of default based on the quantitative and qualitative factors.
These three credit risk components are used to assess the credit risk associated with both
borrowers and instruments, which is reflected in credit scores. The traditional approaches to
credit analysis involve examination of the above five C’s, subjectively weight them, arrive at
a credit score and reach a credit decision (Figure-1).

Figure 1: Credit Appraisal and Scoring Process

Assessment of the above parameters is usually made on the basis of traditional data sources
such as credit history reflecting the existing financial footprints of prospective borrowers as
a proxy for the character and capacity. Further, other financial data such as net worth, tax
payments, income stream, financial assets, etc. are used for an assessment of the other
parameters. Almost all banks use this system for their credit appraisal and sanction process,
however, this has not been able to prevent the occurrences of loan defaults and breach
of covenants and

2
secondly, the problem with an approach based on the existing financial footprints is that it
excludes the hitherto unbanked and under-banked segments of the economy both in the
personal and business segments. Moreover, many times the financial statements fail to
present a true and fair position of the business and the projections may not be realistic. As
such, there are certain inherent problems with this approach, summarized as under.

1. Non-existing financial footprints: For the hitherto excluded segments

comprising individuals, unorganized and informal sector, and underserved segments
such as MSME, this presents a serious challenge to the traditional approach to the
credit appraisal as the required financial footprints in form of credit history, financial
statements, tax returns, financial and real assets may not be available.

2. Disguised financial statements: In cases where the financial statements and

projections are available, there is always a possibility of these being camouflaged or
exaggerated and hence, any assessment made on the basis of such information is
bound to be unrealistic.

3. Inconsistency: It is difficult to apply the credit information parameters uniformly

and consistently across different borrowers and therefore, appraisals may not remain
consistent over a cross-section of borrowers.

4. Subjectivity: There is always a scope for subjective bias in finding the optimal
weights to apply to the parameters while assigning the credit score. This makes
comparability of rankings and decisions very difficult for a neutral observer
monitoring a credit analyst’s decision and for the peer group in general.

5. Myopic appraisals: Appraisals tend to be myopic and lack a futuristic view due to
the inability of the traditional methods to have a foresighted assessment. Since there is
hardly any real-time data, there remains limitation in using the historical data as a
proxy for the future performance.

6. Lost business potential: Finally, the inability to extend credit to certain segments
due to failure of the appraisal methods and also the possibility of an adverse selection
in other cases is a serious challenge for credit managers.

3
 III. CREDIT GAP, BUSINESS POTENTIAL AND THE NEED FOR
ALTERNATIVE APPROACHES
The extent of financial exclusion is quite deep in the most parts of the world in the retail,
unorganized and loosely organized sectors. While there is lack of robust and updated data on
the extent of exclusion for the individual or the unorganized segments, some reliable data is
available regarding the MSME sector, a majority of which remains in the unorganized or the
loosely organized segment. The SME credit gap has proven to be an enduring structural
feature across both developing and developed markets. In the world’s developing markets,
about half of the estimated 400 million SMEs, or 180 to 220 million SMEs, still have unmet
credit needs totalling US$2.1 to US$2.6 trillion1. Only 5% SMEs were fully served in terms
of the availability of credit (Figure 2). The credit gap results from both demand and supply-
side problems. Many SMEs are usually found reluctant to seek or unable to access credit due
to the documentation and collateral requirements for obtaining a loan, high costs, long
decision hierarchy and timeline. Other the other hand, many banks consider SMEs/MSMEs
to be high-risk and high-cost clients to acquire and service.

Figure 2: Global SME Credit Gap (Source - GPFI)

In case of India, according to MSME Ministry’s Annual Report for 2016-17, MSME sector in
India boasts of 51 million enterprises providing employment to 117.1 million persons and
contributing about 33% of the manufacturing output and 45% of the

4
export output. However, as per the provisional data for period ended March 2016, total
outstanding loan of the banking system to MSME sector stood at around 11.1 trillion rupees
in 20.6 million loan accounts as against the estimated need of INR 26 trillion and number of
MSMEs at 51 million2. This shows the extent of credit gap and consequently the untapped
business potential for banks. It makes sense both from the policy and business perspectives to
find ways of bridging the credit gap in a manner which is both enduring and remunerative for
the financial market participants. Alternative approaches to credit dispensation provide a
credible solution in this direction.

IV. THE ALTERNATIVE APPROACHES TO CREDIT RISK AND THE USE

OF DATA
A sound process of the credit appraisal must capture the ongoing transaction risk, exposure
and concentration risk while adequately focussing upon the intrinsic default risk. While the
alternative approaches to the credit scoring can successfully work with some of the hitherto
under-served segments of the market, these could also act as an additional layer of credit
analysis for other market segments to supplement the traditional models.

It is estimated that approximately 2.5 quintillion bytes of data are produced in the world
every day3. At the end of 2016, the world had a total of 4.8 billion unique mobile
subscribers. By 2020, almost 860 million new subscribers will be added, taking the global
mobile phone penetration rate to 73%. Ten countries are expected to account for 72% of
growth in new mobile subscribers worldwide. India, already the world’s second-largest
mobile users market is bound to be the primary driver of this growth, with 310 million new
unique subscribers expected up to 2020, helped by improving affordability, falling device
prices and better network coverage4. Consequently, good quality data can be drawn from
continually expanding sources, namely, electronic transactions, mobile call records, talk-time
purchase patterns, credit bureau information, social media posts, geospatial data, etc. These
emerging sources of data have the capacity to positively impact the quality of credit
appraisal.

As discussed, the appraisals made on the basis of repayment history and existing debt have
their own limitations inherent in every point-in-time method based on the

5
historical data. However, in today’s digital age the prospective borrowers continually
generate huge quantum of behavioural, transactional and social data that can provide robust
indicators so as to assess their creditworthiness and repayment capacity. The alternative data
can mean anything and everything beyond the traditional credit history usually gathered by
banks and credit bureaus. Some non- financial data sources have garnered
particular attention namely, social media data,
demographic data, transactional data relating to
the payment history and utility bills. These are PsychometricSocial Media
gradually emerging as very useful input for the DataData

estimation of credit scores for clients who may be

Consumer
otherwise unserviceable. The psychometric Demographic
Transaction Data
Data
analysis is also gathering
acceptance as an alternative method of credit Figure 3: Alternative Sources of Data
appraisal (Figure - 3).

Further, a large proportion of the newly available data is passively produced as a result of
people’s interactions with digital services such as mobile phones, internet searches, online
purchases, and electronic payment transactions. Characteristics of individuals can be inferred
by developing algorithms that make use of these data. As the primary generators of the data
are not in the control of their digital footprints and also cannot guide the ways in which it
may be used, there is hardly any risk of manipulated or camouflaged information as in the
case of the traditional financial statements and projections. Though there remain concerns
about the privacy, the access to high-quality unadulterated data and ability to make decisions
based on such data is a great accomplishment in the challenging world of credit appraisal.

A large numbers of low-income people, micro-entrepreneurs, small-scale businesses, and

rural populations that previously did not have access to formal financial services are now
increasingly being digitally banked by a range of old and new financial services providers
including non-traditional providers such as mobile network operators, NBFCs and FinTechs.
Alternative credit market has emerged across the globe as reflected in Figures – 4, and 5

6
Figure 4: Alternative Credit Market (UK, USA, China)

Figure 5: Alternative Credit Market [Canada, Latin America, Australia, Asia Pacific (excl. China), Europe (excl. UK)]

It can be observed from the above figures that the alternative financing to the SME segments
have expanded across the globe, albeit with varying growth rates. Highest growth of P2P
lending is seen in case of China and the Asia Pacific due to greater credit gaps and inability
of the traditional financial system to meet the requirements. In other developed markets, the
balance sheet lending and invoice financing have also picked up beside the P2P market.

7
 V. RECENT DEVELOPMENTS IN ALTERNATIVE CREDIT SCORING
Following are some of the success stories from across the world which show the
immense potential of the alternative credit scoring and its positive impacts.

 A study conducted by researchers at the Massachusetts Institute of Technology (MIT) has

demonstrated that the introduction of technology-enabled financial services can help
reduce poverty5. The study estimated that since 2008, access to mobile money services
that allowed users to store and exchange money increased daily per capita consumption
levels for roughly two percent of Kenyan households.

 FICO Score XD from LexisNexis Risk Solutions6 and Link2Credit First Score Direct7 are
the models which target the potential borrowers in the USA with little or no credit history.
These models use aggregated public or proprietary data to develop an alternative risk-
based approach for credit analysis. Their data sources typically include employment and
address history, property and asset ownership, bankruptcies, liens and decrees, criminal
records, identity records, memberships, utility data, etc.

 Commercial Bank of Africa (CBA) and mobile operator Safaricom launched M- Shwari8,
a digital savings and loan product that has provided small credit limits over mobile phones
called nano-loans to millions of borrowers, bringing them into the formal financial sector.

 Tiaxa turn-key nano-lending approach9 builds on the GSM usage, payroll, utility
payments, money transfers, etc. as benchmarks for assessing the creditworthiness of the
prospective borrowers. Tiaxa uses a range of machine learning methods to reduce
hundreds of potential predictors into an optimal model. Custom models are designed for
each engagement. Tiaxa now has more than 60 installations, with 28 clients, in 20
countries, in 11 MNO groups, who have over 1.5 billion end users among them.

 Lenddo10, a Singapore based company, is another example which has been successful in
rolling out social media credit assessment across multiple geographies such as the
Philippines, Colombia and Brazil. Leveraging a proprietary algorithm, Lenddo rates
borrowers on a scale of 1 to 1000 based on

8
 their likelihood to repay a loan. Lenddo loan applicants give permission to access data
stored on their mobile phones as the phone holds thousands of data points that reflect
one’s personal behaviours, such as social connections, activity, group memberships, and
communications. The applicant’s raw data are accessed, extracted and scored, and later
destroyed by the Lenddo. More than 50 elements across all social media profiles provide
12,000 data points per average user which are harnessed to gather a reliable and useful
inference about the creditworthiness.

 Vodacom11 a mobile service provider in Tanzania, has partnered with First Access12 a for-
profit social business focussed on data appraisal using prepaid mobile data to predict credit
risk for consumers who have never had a bank account or a credit score. First Access
offers an instant risk scoring tool for low- income customers by leveraging demographic,
geographic, financial and social network data from a subscriber’s mobile records. The
solution offers a credit risk rating in real time along with a recommendation on the loan
size and eligibility for instant disbursal.

VI. BEHAVIOURAL ESTIMATIONS – PSYCHOMETRIC CREDIT SCORING

Michael Pompian (2012)13 researched and advanced the science of human attitudes towards
investing and concluded that psychology and the study of the human behaviour can deliver a
better understanding of motives, desires and drivers relating to the financial decisions. By
categorising people into approximate types, he postulated that each individual type should
have a corresponding portfolio suited to one’s temperament, loss aversion, and risk
appetite. Davies & Haisley (2013)14 further extended the concept of behavioural finance
to include a utility calculation that measures an individual’s emotional personality in
respect of anxieties about financial losses versus desires to make financial gains. Measured
by outcomes, the utility position of an individual means the chances of a more refined,
better-attuned portfolio management. Taking the basic algorithms of the behavioural
finance and coupling them with the complex correlations of predictive appraisal can
provide real observations that can be translated into valuable insights about risk profiling
and estimation of the creditworthiness. Using the concept, psychometric surveys that use

9
a set of questions to evaluate a potential borrower’s ability and willingness to pay are
becoming gradually popular as a credit risk assessment tool.

Entrepreneurial Finance Lab (EFL)15 has pioneered in the psychometric credit scoring
through research at the Harvard Center for International Development. It leverages
psychometrics to evaluate the creditworthiness of borrowers in over 20 emerging countries,
including India. Recently, Lenddo, EFL and Orient Commercial Bank have joined together to
serve the unbanked population of Vietnam16. The psychometric test delivered in vernacular
languages is a 30-45 minute survey that includes suitable controls for fraud and gaming.
Applicants are assigned a three-digit credit score that predicts a borrower’s probability of
default17. EFL entered the Indian market in 2013 and has joined hands with several NBFCs
claiming that lenders using its screening tool have shown up to a 50% reduction in the default
rate. Though it may be too early to comment on the efficacy of their assessment, the initial
results appear promising.

VII. CONCLUSION
Alternative credit scoring is increasingly gaining ground to cater to the credit needs of the
under-served segments lacking a sufficient credit history. In addition, this can also offer a
second level appraisal support for other segments of the credit market. Establishing an
appropriate credit risk environment, operating under a sound credit granting process,
maintaining an appropriate credit administration, measurement and monitoring process and
ensuring adequate controls over credit risk are the series of inter-related and inter-dependent
steps for ensuring successful management of credit risk. Sound credit scoring techniques
play a significant role in this process.

Indians are the second-largest mobile phone users in the world 18 and this provides a huge
potential for the use of alternative credit scoring. Every time individuals make a phone call,
send a text, browse the Internet, engage social media networks on their phones, or top up
their prepaid cards, they deepen their digital footprints. It is possible to undertake a digital
risk profiling in order to determine the creditworthiness of underserved customers. Ability to
analyse non-linear incongruent information streams has immense potential to boost the
scope and credibility of such credit

1
scoring models. Psychometric testing coupled with the data based alternative credit scoring
methods provide added advantages in developing robust credit scores based entirely on the
alternative data. The availability and predictive power of the alternative data sources vary
widely, and this may suit needs for different lenders in various markets. However, the credit
analysis and scoring based on alternative data or otherwise is only one component of the
lending due-diligence process and therefore simply a good credit score cannot guarantee a
strong credit performance. Finally, in certain cases, the alternative credit scores may be used
as a supplement rather than a substitute to provide a more nuanced understanding of credit
risk and the business potential. Finally, the alternative credit scoring can act as a very useful
bridge between digital footprints and financial footprints and thereby can help both the
potential borrowers and financial institutions to achieve their respective goals.

1
REFERENCES

1
Global Partnership for Financial Inclusion, Alternative data transforming SME finance, May 2017,
https://www.gpfi.org/sites/default/files/documents/GPFI%20Report%20Alternative%20Data%20Transf orming
%20SME%20Finance.pdf
2
ABCD of MSME credit, https://www.rbi.org.in/Scripts/BS_SpeechesView.aspx?Id=1018
3
‘The 4 Vs of Big Data’, IBM Big Data Hub, accessed April 3, 2017, https://www 01.ibm.com/ software/data/
bigdata/what-is-big-data.html
4
The Mobile Economy 2017, https://www.gsma.com/mobileeconomy/
5
Suri and Jack, ‘The Long Run Poverty and Gender Impacts of Mobile Money’, Science Vol. 354, Issue 6317
(2015): 1288-1292
6
FICO Score XD is a consumer reporting agency product provided by LexisNexis Risk Solutions Bureau LLC
and may only be accessed in compliance with the Fair Credit Reporting Act, -
https://risk.lexisnexis.com/products/fico-score-xd
7
https://www.businesswire.com/news/home/20080103005125/en/L2C-Upgrades-Score-DirectService
8
Data appraisal and digital financial services, IFC Handbook, June 2017, www.ifc.org
9
ibid.
10
http://www.forbes.com/sites/tomgroenfeldt/2015/01/29/lenddo-creates-credit-scores-using-social- media/
#599e2fa73f79; https://www.lenddo.com/
11
http://www.vodacom.com/about-us/home
12
http://www.firstaccessmarket.com/
13
Pompian, Michael M (2012), Behavioural Finance and Investor Types
14
Davies, G. B., Haisley, E (2013), Overcoming the cost of being human (or, The pursuit of anxiety- adjusted
returns
15
https://www.eflglobal.com/about/
16
https://include1billion.com/news (25 January 2018)
17
http://businesswireindia.com/news/news-details/addressing-missing-middle-efls-psychometric-credit-
scores-help-financial-institutions-lendentrepreneurs/38558;http://social.yourstory.com/
2014/01/minority-report-efl-stop-loan-defaulters/
18
Internet and Mobile Association of India (IAMAI) research; ‘Digital, Social and Mobile in India in
2015’ –We are Social

1
Author’s Profile
Dr Ashish Srivastava is an Assistant General Manager and a Member of Faculty in the Co-
operative Banking Channel at the College of Agricultural Banking, Reserve Bank of India,
Pune. He holds a PhD in Working Capital Management, and Master of Business
Administration (MBA) with specialisation in Finance. He has obtained the degree of Master
of Finance from the Judge Business School, University of Cambridge, UK and has also
qualified the eligibility test for university lectureship in India. His qualifications include the
Certified Associate (CAIIB) and Certificate in Treasury and Risk Management (CTRM) of
the Indian Institute of Bankers. He also possesses the Financial Risk Manager (FRM)
certification along with the International Certificate in Banking Risk and Regulation
(ICBRR) of the Global Association of Risk Professionals (GARP), USA. In addition, he is
accredited with the Bloomberg Market Concepts Certification.

He has about twenty years’ experience in the field of banking and finance and has taught
postgraduate courses in business management, besides having an extensive experience in the
area of banking supervision. Prior to joining the Reserve Bank, he has worked with the Food
Corporation of India in a managerial capacity in the area of finance & accounts. His
research interests include banking regulation, systemic risk and financial innovations and he
has published more than 50 papers in the leading journals and magazines. His areas of
specialization include statutory inspections, risk management, credit management, investment
management, capital adequacy, regulatory compliance, macroeconomic issues, etc. He could
be reached at [email protected]

The views are personal and do not necessarily reflect those of the Reserve Bank of India.