Page 208 - ITU-T Focus Group Digital Financial Services – Technology, innovation and competition

ITU-T Focus Group Digital Financial Services
Technology, Innovation and Competition

E ACCESS TO AND USE OF BIG DATA SETS

10 Big data and DFS

10.1 Overview

As DFS evolves from its genesis as primarily a remittance-type service to a more transactional offering that
includes services such as insurance, investments and credit provision, SPs may want better data sets to assist
them to develop new products, to assess customer risk, and to target the correct market segments.
186
For provision of credit, be it short-term micro-credit or a longer term macro-credit product, providers need
specific data sets to assess risk and credit worthiness. The data is limited though: only 10% of people in eight
187
sub-Saharan countries, for example, have verifiable online financial data.
188
For many DFS markets, the most cogent data sets are often those that can be gleaned from mobile phone
use, either from conventional telecommunications activity use, through transactional data in DFS or similar
transactions obtained by DFS providers such as MNOs, or through third party smartphone app providers.
189
In the telecommunications (use) context for example, Call Data Records (CDRs) captured in the course of
their operations by MNOs are evolving from simply being flat records of telecommunications service use by
individual customers to being the cradle of rich data insights made possible by the connective tissue of big data
algorithms. This so-called ‘exhaust’ data scrapped from these data sources can reveal a lot more on customer
behavior, and thus credit worthiness. These metrics are the maximum types of data sets that can be derived
190
from customers with feature phones, augmented however if the MNO also provides DFS products.
191
Even richer data sets can be gleaned from users with smartphones, who may use apps that reveal further
information about them. For example, some new smartphone apps from DFS credit providers will request
and obtain from the user consent to mine their contact lists, get device details, obtain biographical data in
registration forms beyond that can be obtained in (often mandatory) SIM card registration, as well as track
their calls, SMSs, instant messages, digital purchase habits, and location. Similar data and results can be
192
obtained by messaging and social network apps who have payment components added, such as those from
Tencent’s ‘WeChat Pay’ application in China, and social network behemoth Facebook’s ‘Messenger’ application.

This accumulated data becomes valuable in creating alternate credit scores and in then facilitating provision of
credit to some of these profiled users. In many cases, however, users may not be aware data is being scrapped

186 See further on the nature of adverse selection and data sets, Mazer & Rowan (2016) ibid; and generally on big data and DFS,
Chen, G & Faz, X (2015) The Potential of Digital Data: How Far Can It Advance Financial Inclusion?, available at https:// goo. gl/
dxxSIU
187 This information asymmetry, in a credit-provision context, may result in what is termed adverse selection, such that without a
credit risk assessment – or credit score – the borrower will seek and often be given credit by lenders who are unable to obtain
enough information on hand to have made a more seasoned determination of whether the loan would be repaid. Thus, those
with access to cogent data sets will mitigate the risks of adverse selection. See further Mazer & Rowan (2016) ibid
188 Christensen P (2015) Credit Where Credit Is Due, available at https:// goo. gl/ h0Oapm
189 Data can of course be gleaned from bank-related activity but this may be restricted through bank secrecy laws in some countries
- for example Pakistan - which have often gotten in the way of sharing data that could otherwise be valuable in the hands of
alternative financial providers. Here then, traditional credit providers benefit from their ‘proprietary’ data.
190 See San Pedro, J et al (2015) MobiScore: Towards Universal Credit Scoring from Mobile Phone Data, available from https:// goo. gl/
Mkwp5T
191 Also through some feature phones that have Facebook, Twitter, and Whatsapp installed. See further, Perlman, L (2017) DFS
Handset Overview: ITU FG on DFS, available at http:// www. itu. int/ en/ ITU- T/ focusgroups/ dfs/ Pages/ default. aspx
192 In most cases prospective (and existing) users can only install and thus the app to get credit only if they agree to all these metrics
being monitored.

184