M ICROFINANCE C REDIT R ISK M ANAGEMENT T OOL G UIDES

3 C REDIT S CORING

What is a Credit Score?

 Scoring is a method of assigning a numerical value (the “score”) to a client in order to predict how
likely he or she is relative to others to experience some event or perform some action in the future.
This is predicated on the notion that past behavior is indicative of future behavior for populations
with similar characteristics.
 By analyzing a sample of historical client and business data, trends are deduced to better understand
(potential) clients and predict future events such as credit repayment.

Learn from the Anticipate the

Understand the Present
Past Future

 This tool guide focuses on credit scores – a number that represents an assessment of the
creditworthiness of a person, or the likelihood that the person will repay a loan. Financial institutions
use scoring models to assess the credit risk of a borrower and aid in the credit evaluation processes.
 A score can be applied along the steps of the microfinance lending methodology, providing objective
inputs to make the process more effective and enhance standards and controls. Credit scoring
models are not intended to replace loan officers and other commercial staff, but rather to
complement and facilitate their work by supporting assessment of willingness to pay.

Score Inputs
Scoring modules analyze client data to identify trends that can help predict future behavior. There are many data
sources which a score can use as input:
 Socio-demographic data such as client gender, age, civil status, business experience, time
residing at current residence, type of residence, etc.
 Behavioral data which reflects the history of a client with an institution and may include variables
such as: number of days past due at last month end, maximum number of days a client was
delinquent, average number of days a client was delinquent, number of times client was more than
30 days delinquent; number of times client was more than one day delinquent, etc.
 Credit Bureau data which is collected by a third party and includes personal credit history with
financial institutions (and sometimes phone/utility companies).
In recent years, microfinance institutions in collaboration with third parties have also been exploring newer,
additional data sources to include in risk management tools like scoring models:
 Psychometric data comes from specifically designed questionnaires to determine a person’s
knowledge, abilities, attitudes, and personality traits. In microfinance, these principles can be used
for credit scoring as well as to assess entrepreneurial aptitudes.
 Big data includes data from various sources outside of the institution, such as phone companies,
utilities, retailers or social networks (Facebook, Twitter, LinkedIn, Yahoo!).
Types of Scores
Scores are characterized by the data source as well as the stage in the credit process to which they are applied, the
degree of tailoring to an institution, and the technique used to generate the score.
 Stage in the credit process (see also Applying Scores in the Credit Process below):
̶ Selection Score is used by microfinance institutions (MFIs) to aid in the promotion to and
evaluation of new credit clients, and often uses socio-demographic or psychometric data. Once
new client data has been collected by the loan officer, a Selection Score can be generated in order
to obtain the client credit risk level (e.g., high, medium or low risk). Depending on the client risk
level, the MFI can choose appropriate client strategies such as more or less intensive credit
evaluation procedures or approval conditions.
̶ Collection Score is used to establish the probability that an existing client will repay his/her
credit and can be used to tailor overdue collection tactics. Based on the Collection Score result,
an MFI can group their clients by repayment willingness and then generate focused overdue
collection strategies. For example, a client with a high likelihood of repayment may receive a
reminder phone call if one day past due on a payment whereas a client with a low repayment
probability may require an in-person visit. A Collection Score provides MFI staff responsible for
recoveries with a clear focus and direction for their efforts, leading to increased efficiency and
reduced expenses.
̶ Renewal Score is used to optimize the client credit renewal process and is based on the client’s
previous payment behavior with the MFI. Renewal strategies can be tailored according to the
client risk level, ranging from simple and streamlined approvals for low risk clients to more
complex evaluations and approval processes for high risk clients. Furthermore, an MFI can use
Renewal Scores to segment clients by risk levels and develop targeted strategies to promote
client loyalty and retention – for example, a low risk client may receive preferential pricing when
renewing a loan.
 Degree of tailoring to an institution – A score can be considered generic or tailored depending
on whether it is adapted to a particular institution:
̶ Generic score refers to a score developed for a particular product, client segment, or region,
but not based on specific institutional data. Generic scores are typically used for new institutions
or new products, before enough information is available to develop a tailored score.
̶ Tailored score refers to a score which has been constructed using the specific data of the
institution to which it is applied. It is recommended to have at least two years of historical
information to create a tailored score.
 Scoring technique – There are two main techniques for generating a score:
̶ Statistical Score uses mathematical models and statistical software (e.g., SPSS) to identify
patterns in the data. The methods used most frequently are logistic regression, decision trees
and neural networks.
̶ Expert Score utilizes the knowledge and past experience of “experts” to predict future client
performance.

Credit Scoring Tool Guide -2- www.accion.org

Applying Scores in the Credit Process
In a competitive and changing environment, MFIs need to be efficient in all processes, reduce costs, and manage
credit risk if they want to survive and thrive in the long term. Research in relatively mature microfinance markets
like Bolivia and Colombia has shown that the use of credit scores can significantly reduce an MFI’s operational
costs.
There are many different applications for credit scores and complementary tools along the stages of the credit
process as illustrated in the diagram below (the scores are shown in orange boxes):
 Promotion: A Selection Score is used to aid in the selection of new loan clients.
 Credit Bureau Checking: A Bureau Score is applied based on borrower data available in the
credit bureau.
 Approval: Findings from the Selection and Bureau Scores provide valuable inputs into the credit
approval decision.
 Collection: Collection Scores can be used to improve processes related to existing client collections.
 Renewal: Renewal Scores can be used to improve processes related to existing client loan renewals.

Diagram: Credit Scores in the Credit Process

Tool Benefits
An MFI can gain multiple benefits from implementing a credit score. Most important among them:
 Quality control: Scores assess client quality factors and provide alerts for those clients who don’t
fall within expected norms.

Credit Scoring Tool Guide -3- www.accion.org

  Consistency in decision-making: When using scores, an MFI doesn’t need to depend exclusively
on loan officer’s perception when evaluating a potential client. Decisions are made within a standard
framework or parameters set by the institution, increasing consistency.
 Process optimization and control: Greater transparency in the credit evaluation and overdue
collections processes as a result of score usage makes it easier for the institution to oversee and
manage the actions of the loan officers.
 Regular follow-up and client satisfaction: Client segmentation via scores allows for ongoing
client follow-up, leading to better customer service and reduction in client desertion.
 Improvement of MFI performance: The use of scores facilitates the improvement of key
processes and is expected to result in better overall performance, including:
̶ Reduction in the portfolio at risk (PAR)
̶ Reduction in the turnaround time for credit delivery
̶ Increase in loan officer productivity (new and renewal credits disbursed per loan officer per
month)
̶ Improved overdue collections results (focused efforts allowing for greater staff efficiency and
capacity)
̶ Increased client retention as result of better client service and tailored strategies to reward good
clients

When to Use
Usually an MFI will decide to use scoring techniques by developing a business case and comparing the potential
benefits (e.g. efficiency improvement in a particular process) with the required effort and cost for implementation.
The MFI will also need to verify that the requirements for implementing a score are in place (see below). The type of
score selected and timing of implementation will depend on institutional required process improvements, available
data, and available resources. Many times, MFIs will begin with a renewal score (which leverages existing client
data), and then implement a collection and selection score. To maximize the investment, scores are best applied to
products that comprise a large portion of the institution’s client base or portfolio.

Tool Usage and Users

 The Chief Risk Officer (or risk manager) is the owner of the score and uses it for credit risk
monitoring and management.
 Loan officers and/or branch managers use client score results as input for strategies to work
with (potential) clients and assist in the various steps of the credit process.
 Senior management (MD/CEO, Chief Commercial, Operations and Financial Officers, heads of
departments) can use scoring tools to better understand the institution’s client profile and to
optimize the MFI’s processes. For example, for strategic reasons senior management may decide
they are willing to take on additional risk to increase client approval rates by adjusting cut off rates
for the score parameters.
Given the complexity of scoring, development and implementation is often handled by a specialized external
consulting group in close collaboration with the senior management team and the risk department.

Credit Scoring Tool Guide -4- www.accion.org

Tool Requirements
The implementation of a score in a microfinance institution is a complex and challenging initiative. Success depends
as much on effective score implementation and monitoring, as it does on the correct development of the score.
Institutions considering score implementation should make sure the following required conditions are in place:
 Strong willingness of senior management to implement this tool including commitment of the
necessary resources and attention
 Access to sufficient data for the score – either socio-demographic, behavioral, or credit bureau
 Accurate portfolio database with historical information about each client
 Implementation team and plan for score rollout, starting with a pilot in select branches and then
expanding to the whole institution
 Technology unit with the capacity to add and modify the score in the core banking system without
disrupting the functioning of the system and the institution
 Score monitoring and tracking tool to assess whether the score is working correctly or not (proper
monitoring will also require regular follow-up and analysis by a scoring expert).
 Strong and transparent credit processes for client selection, evaluation, approval, disbursement,
overdue collections, and renewal
If the above requirements are in place, a score can be developed in a few months – pilot testing, score refinement
and full roll-out can take up to a year or more. Once implemented, the process of score monitoring and refinement
is ongoing.

Tool Results
The following examples illustrate some typical credit score results: 1. Enter client ID
Selection Score
 The image to the right shows a selection score for a new client. The
score is calculated in the system for each new client individually,
once the client data has been entered on the system. The score can
be retrieved by 1. Entering the client ID, 2. Requesting the score to
be calculated, 3. Retrieving the score from the system. In the 2. Press calculate
example, the resulting score is classified as low risk (‘Risco Baixo’).
 The table below illustrates client segmentation by risk level. In this
3. Retreive the score
example, new clients are assigned a particular level of risk and
segmented into risk categories: A+ (best clients, lowest risk) to F
(worst clients, highest risk).

Credit Scoring Tool Guide -5- www.accion.org

 SELECTION SCORE Jan/14 Feb/14 Mar/14 TOTAL
Score for New Clients 664 761 658 2,083
Score A+ 31 38 30 99
Score A 50 55 39 144
Score B+ 77 90 68 235
Score B 83 103 92 378
Score B- 168 188 160 516
Score C+ 71 70 65 206
Score C 133 153 143 429
Score C- 51 64 61 176
Score F --

Renewal Score
 In this example, clients with loans that are soon to mature are listed with client ID, information on
their previous credit history with the institution and a recommended renewal evaluation
classification, process and loan amount. This list can be used to prioritize clients for renewals and
ensure a suitable evaluation process for the client risk level is applied:
Recommended Maximum
Previous Loan Amount Recomme
loan Loan as % of nded
Client amount Loan Closure Recommended Previous Loan Loan LO
ID ($) # Date Classification Process Amount Amount Name
15 10,000 8 13-Sep Detailed Business visit, 90% 9,000 Joe
update client
information
83 5,000 3 3-Sep Pre- Automatic 150% 7,500 Paula
approved approval, no
need to update
client information
98 8,000 9 13-Sep Reject Reject the client - - Brad
138 8,000 8 18-Sep Reject Reject the client - - Tom
155 7,000 10 2-Sep Full Business and 75% 5,250 Angie
home visit,
update client
information
158 5,000 3 10-Sep Pre- Automatic 150% 7,500 Jackie
approved approval, no
need to update
client information

Credit Scoring Tool Guide -6- www.accion.org

 Collection Score
 In this example, existing clients (identified by client ID) are listed along with information on their past
due payment(s) and a corresponding recommended overdue collection strategy (e.g. no action, phone
call, staff visits with notification letters type A, B or C). Loan officers (and other commercial staff) use
this list to prioritize their collection efforts and apply the appropriate overdue collection strategy.

Date of Client Loan

Execution ID No. Days Past Due Overdue Collection Strategy* Loan Officer
31-Aug 22 XY5672 18 LO, BM visit (Notification C) Kwame
31-Aug 88 XY8760 20 LO, BM visit (Notification C) Kwame
31-Aug 45 XY3642 11 LO, BM visit (Notification B) Kwame
31-Aug 90 XY5153 7 Phone call Kwame
31-Aug 44 XY2945 3 No action Kwame
31-Aug 67 XY1639 7 LO, BM visit (Notification A) Kwame
* BM = Branch Manager, LO = Loan Officer

Scoring Examples for Financial Inclusion

As alternative data sources and underwriting methodologies become more and more prominent, the importance of
credit risk tools like scoring is expected to become more widespread. A few recent, innovative examples are outlined
below:
 Scores based on online/social media data: Some start-up companies are using social networks
like Facebook, LinkedIn, and Twitter to estimate creditworthiness. These companies look at a
number of data points – where you’re located, who you’re connected to, and how transparent you are
– to build a ‘credit score.’ Examples of companies leveraging this online data include DemystData,
Lenddo, and Zest Finance.
 Scores based on cell-phone usage: Other companies combine financial and mobile data to
reliably predict credit risk for borrowers in informal markets. Cell phone records are analyzed to
assess demographic information, geographic information, strength of social network, and financial
information, all of which can be considered a part of client evaluation. Examples of companies using
this data include Cignifi, First Access, Mines.io and Tiaxa.
 Scores based on psychometric testing: Psychometric tests include questions relating to a
person’s character, abilities, and attitude. These tests scores, combined with analytics and modeling
techniques, can deliver a credit score identifying high-potential, low-risk entrepreneurs. Examples of
companies using psychometric testing include the Entrepreneurial Finance Lab and VisualDNA.

Tool Variations
 Different types of scores can be implemented in a MFI according to its maturity, number of products,
geographic diversity and other specific characteristics.
 MFIs may start with an expert or generic score and then transition towards a statistical or tailored score.

Credit Scoring Tool Guide -7- www.accion.org

Complementary Set of Tools
The following additional tools can be used to complement scores in the credit process:
 Economic activity alerts: compare the economic activity and sales volume of the client business
with reference parameters and provides alerts about whether a microenterprise’s values are outside
of the normal parameters.
 Smart evaluation tool: performs an automated analysis of the microenterprise and the micro-
entrepreneur. It performs part of the analysis that the credit analyst typically carries out in the credit
process, using scores and the economic activity alerts as inputs.

Credit Scoring Case Study 1: Use of a Client Selection Score to Increase Loan Officer Productivity for Small Working
Capital Loans

Institution First Caribbean MFI is a leading microfinance institution in the Caribbean that has been in existence for
more than 10 years. It has a regional branch network in all major urban and peri-urban areas, a portfolio
of approximately $10 million and over 10,000 clients. Its main loan product is working capital loans for
micro and small businesses (average loan size $950).

The First Caribbean MFI was facing the following challenges:

Problem  Low loan officer productivity
 Stagnant portfolio growth with disbursements only sufficiently compensating for loans closed
 Small loan amounts ($400-$500 on average) at low productivity were not enough to compensate for
the operational cost of the underwriting process

Solution/ To address the challenges noted above, First Caribbean MFI introduced a Client Selection Score to
Score Used segment and predict the best potential clients for their small working capital loan product. The score is
based on socio-demographic characteristics (e.g. client age, time spent in the business, marital status,
level of education, house ownership, business location stability). The best potential clients as
determined by the Client Selection Score are eligible for a streamlined, faster process for credit
evaluation and loan approval, increasing loan officer productivity and promoting growth.

LO Productivity (new loans) Number of Active Clients

25 16,000
14,000
20 12,000
10,000
15
8,000
10 6,000
4,000
5
2,000
- -
Ma…

Sept
Nov

May

Aug
Sep
Oct

Dec
Jan
Feb

Oct
Nov
April

June

Dec
July

Sept
Sep
Oct

Oct
Nov
Dec
Jan
Feb
Mar
Apr
May

Aug

Nov
Dec
Jun
July

Credit Scoring Tool Guide -8- www.accion.org

 Credit Scoring Case Study 2: Use of a Renewal Score in Evaluation Client Loan Renewals

Institution Peace Microfinance Bank is a small bank offering microfinance products (savings and loans) in Africa.
After two years of operations, Peace had reached a loan portfolio of 8,500 clients and more than $5
million. The main loan product is a working capital loan for small businesses (loan size $700-1,200).

The Peace Microfinance Bank was facing the following challenges:

Problem  Complex underwriting process for existing clients that required more resources than for new clients
 Low new loan productivity due to excessive dedication to renewals (only five new loans/LO per
month)
 Renewal ratio low (at 71%when business plans required at least 75%)
 High PAR 30+ (at 14% and increasing)
 High renewal loan PAR 30+ (at 16%)

Solution/ To address the challenges noted above, ABC MB introduced a behavioral Renewal Score, allowing the
Score Used automated segmentation of the current portfolio by client risk level according to historical client
behavioral data.
Main data used in the score included variables such as average number of delinquency days during
the last quarter, number of days late in last loan term, number of months as client of the bank,
maximum number of delinquency days in last quarter, etc.
The tool is used to identify good clients and provide them with enhanced services – for example, an
offer of new products or the same product with better conditions.
Achieve- As a result of introducing the Renewal Score, Peace Microfinance Bank obtained the following benefits six
ment/ months after score implementation:
Benefits  Renewal productivity (number of renewed accounts per loan officer per month) more than doubled
– grew to an average of 14 per LO per month
 Portfolio at risk (PAR 30+) for the renewal portfolio came down from 17% on average to 9%

Renewals / LO per Month Renewals PAR (30+)

18 20%
16
14 15%
12
10
10%
8
6
4 5%
2
- 0%
Mar

Apr

May

June
Aug

Oct

Nov
Sep

Dec

Jan

Feb

Mar

Apr

May

June
Aug

Oct

Nov
Sep

Dec

Jan

Feb

A version of this tool guide is also available as an interactive online module– you can access it by visiting:
http://accion.weejee.com/module2/story.html.
The Credit Scoring tool guide is one of a four-part series, which also includes tool guides on Detailed Risk and Controls
Assessment (DRACA), Branch Transaction Risk Reports and Portfolio Quality Analysis (PQA).
If you have any additional questions, or would like support in understanding and implementing this tool guide, please contact
Accion’s Training and Capacity Building team at TCBSupport@accion.org.

Credit Scoring Tool Guide -9- www.accion.org