Originally posted on the World Bank Blogs Platform.

Have you noticed the large number of digital trails that are left behind in your browser and social media? Have you ever received an offer for a product or service that you were just thinking of?  A friend of mine was researching about a critical illness and looking for insurance plans on the internet at the same time, and she started receiving ads to secure “the right spot” in a graveyard!  

Now the question arises if these digital trails are leading to accurate analyses that result in adequate decisions, or not.  Each one of us generates an immeasurable amount of data every day and many businesses (and individuals) analyze this data and generate profits from it. The use of big data has expanded to many sectors, but for the financial industry, the advent and use of big data have been regarded as a boon.  Financial institutions and alternative lenders  are now able to use inexpensive yet efficient tools to process loan applications and design personalized products based on demographics and behavioral analysis.  And the consumers benefit from products (like loans and credit cards) that they didn’t have access to before.

Alternative lenders rely on big data to perform some of their critical operations. The future of lending seems to be moving towards making decisions based on scores populated by the correlation between activities such as a visit to a doctor, buying a brand of grocery or choosing a mobile phone.  These are the new variables used to model credit scores to predict payment behavior. Some of these variables are, real-time transactional data, social media, web browsing, news feeds, blogs, GPS data, images, etc. The interpretation of these variables provides the lenders a hint or a clue about the borrower’s credit behavior.  In addition to these new sources of data, machine learning (ML) and artificial intelligence (AI) are helping lenders overcome the challenge of servicing the previously underserved borrowers and improve their earnings significantly. AI and ML find patterns in data and behaviors that the borrowers didn’t expect would be relevant to their financial access. However, this capability raises questions about the appropriate access, use and sharing of such information.

The development impact of the use of such data in the financial sector cannot be ignored at this juncture of the worldwide technology-led revolution.  Many small businesses and individuals that did not have access to financial services before, can now opt for some kind of financial product.  Simultaneously, the potential discrimination or inconsistencies that may result from using information from many sources without proper validation, in a “black box” environment, is an important challenge in this digital era.  Consumer protection and privacy advocates are concerned about the potential violation of consumer rights from the lack of informed consent by the individuals whose data are being collected, processed and shared.  Furthermore, even where existing laws offer some protection against discriminatory decisioning tools, officials are likely not to have the capacity or resources to address the unique concerns raised by the new digital applications and their unexplored potential. 

While responsible use of big data holds many possibilities for financial inclusion, some regulatory bodies are trying to mitigate the risks by setting up “sandbox” or innovation testing environments to support data innovation while testing the implications of using these alternative data and tools against laws and good practices.  The aim is to find an adequate balance to meet the consumer needs, the financial institutions’ economic expectations and a legal framework that is fair to all parties.  Appropriate use of big data in the financial sector may need to be defined by general principles to guide the legal and regulatory provisions to mitigate the potential risks, and various standard setting bodies have already come together to work on this topic from different angles. The International Committee on Credit Reporting (ICCR), in collaboration with the World Bank Group, has published a report on the  “Use of Alternative Data to Enhance Credit Reporting to Enable Access to Digital Finance”, which provides practical policy recommendations on how countries can adopt and leverage the use of alternative data for credit reporting, while mitigating the risk. Additionally, the standard setting body is currently working on  guidelines for credit scoring approaches. This publication will cover the policy recommendations on credit scoring, encompassing both models and decisions, to help regulators in their oversight roles and to also aid in promoting transparency in using innovative credit scoring tools.