Unleashing the Power of Data for Financial Inclusion
1. Unleashing the Power of
Data for Financial
Inclusion
Financial Inclusion Indaba
July 14, 2017
2. Over the last decade we have moved from a data desert to a world
of data abundance – but there is still a disconnect between the data
and decision making
6. Where are we doing this?
GIS CoP: Tanzania, Kenya,
Uganda, Zambia, Pakistan
Measurement frameworks pilot: Nigeria,
Kenya, Malaysia, Philippines, Mexico, Zimbabwe
In-country communities: Rwanda,
Kenya, Ghana, Uganda, Tanzania,
Senegal, Mozambique, Zambia
TAs: KCB (Kenya),
Airtel (Uganda)
Case Study:
Branch (Tanzania)
GIS Innovation
Grant: Central
Bank of Nigeria
7. Policies must be informed by appropriate measurement
What is the current state?
What is the
problem and what
do I need to do?
What indicators
and what’s an
appropriate
target?
Measure progress
towards goal.
Why are there
deviations from
the goal?
8. A closer look at uptake
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Dormant (0 transactions in a typical month) Mailbox (1 - 2 transactions in a typical month)
Used (3 or more transactions in a typical month)
Source: Findex 2014
% of adults with a bank account
9. Zambia: Are people’s needs met?
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sending over
distance
Life events Bill payments Growth Resilience Receiving
income
Local
payments
Liquidity
Savings Credit Insurance Payments
Source: Zambia FinScope 2015
% of Zambian adults that indicated using specified
financial products to meet each use case
10. Survey Pilots: Needs, Drivers, Usage, Outcomes
Needs
• Resilience,
• Ability to Meet Goals,
• Liquidity,
• Transfer Value (e.g.
receive income, make
payments, local and
int’l remittances)
Drivers
• Functional
• Non-Functional
(Behavioral, Cultural,
etc.)
Usage
• Frequency
• Value
• Recency
• Duration
Outcomes
• Financial Health. Is the
person sufficiently:
• Resilient?
• Able to meet his/her
goals?
• Liquid
12. Where are we doing this?
GIS CoP: Tanzania, Kenya,
Uganda, Zambia, Pakistan
Measurement frameworks pilot: Nigeria,
Kenya, Malaysia, Philippines, Mexico, Zimbabwe
In-country communities: Rwanda,
Kenya, Ghana, Uganda, Tanzania,
Senegal, Mozambique, Zambia
TAs: KCB (Kenya),
Airtel (Uganda)
Case Study:
Branch (Tanzania)
GIS Innovation
Grant: Central
Bank of Nigeria
13. Potential benefits for FSPs
Potential
benefits
Increased
Revenue
Decreased
Costs
Improved
customer
experience
• Enter new markets
• Make more informed decisions
• Organisational efficiency
• Improve scalability
• Get distribution and sales channels
right; reducing churn
• Understand the customer better
and tailor offerings to meet needs
14. Key Barriers to financial inclusion in SADC?
Affordability.
• Income too small/irregular or uncounted for
Identification
• Lack of proof of business address
Accessibility / proximity
• Banks too far & agents illiquid)
Value for money
• Product uptake/ Low usage of products-high charges vs low income
Awareness
• More than 10% adults in Zimbabwe do not use insurance & Mobile Money because they do
not know anything about it
Source: FinScope Consumer Surveys
15. What data sources are available to FSPs to address these
barriers?
Data Sources
Geo-location /GIS
data
Social Media Data
Psychometric Data
Mobile application
data
Sensor / IoT data
Satellite Data
Biometric
identification data
Source: i2i, 2016
16. How are FSPs using these sources of data?
Embracers
Incrementalists
Ground breakers
Collaborators
Innovationinobtainingandanalysingdata
Innovation in types of data used
Airtel Uganda overlays call data
records (CDR) with GIS data to
identify consumer clusters to design
an agent deployment strategy to
reach previously inaccessible
markets
Branch uses alternative data and
new analytical capabilities for credit
scoring to expand access for
underserved and unserved adults in
Tanzania
Kenya Commercial Bank (KCB) used new segmentation
methodology to identify 300 000+ clients working in their
agricultural sector from their existing transactional data to
deliver targeted services that meet farmer needs
Source: i2i, 2016
17. What are the barriers for FSPs to use this data?
Barriers
Uncertainty
around
business case
High initial
investment
Shortage of
required
skillsets drive
up the cost
Sourcing and
aggregating
data is resource
intensive
Fear around
data security
Study in the USA highlighted their need for:
• Evidence on risk assessment
improvements
• Creation of new profitable markets
Lenddo reported that they
incurred significant costs in
the initial data gathering,
development and training of
their algorithm and the
initiative took several years.
The increasing variety
of alternative data
sources means that
data integration and
quality will become a
bigger factor in the
success of new data
and analytical
methodologies
Banks are cautious to outsource
consumer data to specialist firms for
fear that they will be perceived as
taking risks with personal data.
Source: i2i, 2016
Introduce i2i without a slide, all the same intro as the lunch presentation
~302 Billion emails are sent
~2.6 Million blog posts are written
~4.2 Million minutes are spent on Facebook
~984,560 hours of video are uploaded on YouTube
Countries are the 25 World Bank UFA 2020 priority countries
FI = Financial Institution
Interaction:
FI Account: Either a withdrawal or a deposit. Therefore, an account is classified as dormant if the account holder does not make either a withdrawal or a deposit in a typical month.
Statistical universe:
FI Account: The bars represent the % of adults with an account at a formal financial institution. The percentages within the bars, indicate the usage of accounts of those formally served. I.e. the percentage of formally included adults that have a dormant account, mailbox account or use their account.
Additional speaker notes:
While mobile money seems to be used in Uganda, Kenya and Tanzania this seems to be the exception that proves the rule. The Findex Global Survey found that just two percent of adults globally own a mobile money account. However, in Sub-Saharan Africa, the average was much higher at 12%. This represents more than 64 million adults. While this highlights significant penetration of mobile money, the MAP evidence from the MAP countries where mobile money was present showed shows a similar trend to bank accounts: despite high initial up-take in mobile money subscribers, only between a quarter and a third of accounts are active. Active accounts are usually can be defined as those accounts that have been used at least once in the last 90 days.
In Mozambique, mobile money subscribers increased from 485,000 in 2013 to 887, 000 in 2014. However, usage levels have remained low with the proportion of active users estimated to be between 20 – 25%.
In both Swaziland and Malawi, despite rapid initial uptake, the proportion of active users has remained below 30% and the rate of growth of subscribers has slowed substantially. In Malawi from 2013 to 2014 mobile money subscribers grew from about 230,000 subscribers in 2013 to more than one million in 2014. However, less than a third of these adults are active customers.
Similarly, in Lesotho, rapid initial uptake was partially driven by the fact that the providers offered generous incentives - in the form of an airtime bonus - for signing up. Based on in-country consultations, it is estimated that only between one-third and one-quarter of mobile money subscribers in Lesotho are active.
Terminology:
Incrementalists are located in the bottom left of the spectrum and use traditional data sources with long-established methodologies for conventional business decision-making processes. The use of new data sources or analytical processes is more incremental in nature and is applied to improving existing processes or set up as experimental units.
Collaborators are located in the bottom right of the spectrum. Collaborators are traditional players who access new data by partnering with other organisations, typically not in the same sector, to gain additional insight into their customers which they can then use in the design of new services or to reach new markets. The most famous examples are:
Mshwari in Kenya
EcoCash Loans in Zimbabwe
Embracers, located in the top left of the spectrum, recognise the need and power of incorporating new approaches into traditional business practices. They have either adopted new sources of data or new analytical methodologies.
In Kenya, Superfluid Labs enhances their client’s existing customer profiles by combining internal client data with third party data (e.g. credit bureau and MNO data) and public data sets such as social media.
Ground breakers are located in the top right corner of the spectrum. They use both new data sources and analytical methodologies. Many of these organisations are focused on creating new markets and tend to be in the fintech start-up or entrepreneurial space.
Jumo, an African mobile money marketplace provider, uses over 10 000 variables derived from their mobile network and mobile telephone use data, as well as machine learning methods to assess a customer’s credit risk. Data points used include calling records (how many and to whom), airtime usage (amount and top up location), the type of phone used, whether text messages are being sent, data purchases and mobile money transactions.
Branch, a digital credit provider, scrapes an applicant’s mobile phone for thousands of data points, including their contact list, phone make and model, and GPS location when a customer downloads Branch’s mobile application and applies for a loan. Through the use of machine learning algorithms, they then assess creditworthiness and have the applicant’s loan in their M-Pesa account in minutes.
InVenture, whose mobile application scans applicants’ phones for detailed data points, such as the proportion of their contact list with both first and last names, to predict credit risk and provide credit decisions within 20 seconds. Interestingly, InVenture has found that if at least 40% of an applicant’s contact list were organised with both first and last names, they were 16 times more likely to pay on time.