Knowledge Hub

The Impact of Big Data on Improving Digital Finance for Smallholder Agriculture in Africa

First Access is on a mission to turn data into financial opportunity by helping lenders around the world digitize and automate access to credit with powerful, user-friendly technology.

First Access began working on agricultural credit in Africa with the hypothesis that big data could improve the speed and quality of credit decisions for smallholder farmers.

Following on Africa’s success in digital finance, there is hope that predictive, data-driven technology can help unlock more affordable credit. In 2016, the Mastercard Foundation Fund for Rural Prosperity awarded an innovation grant to First Access to develop a comprehensive credit scoring solution for rural borrowers. The innovation would draw on a combination of data from financial institutions and potential partners, and could include crops, market trends, environmental patterns, and transactional and behavioural data of farmers themselves. 

Two years later, First Access has learned a great deal through its research and development, and made a major pivot away from this approach. “A lot of financial institutions are interested in ‘going digital’ and starting with credit scoring,” said Corey Pargee, Director of Operations at First Access. “But we’ve learned from experience that an incremental approach, starting with digitizing the current loan processes, is most effective. Then, a transparent and easily explainable rules engine, segmenting customers based on clear patterns in data, should follow. Adding complex, opaque, or unfamiliar data from external sources is easier and more cost effective after a financial institution has mastered its own data. This includes collecting and analysing the data, identifying opportunities for improvement, and then systematically acting on those insights.” 

The First Access team works with microfinance institutions (MFIs), commercial banks, and alternative lenders that have a mandate to provide credit to low-income rural communities. Competition from digital lenders is driving many incumbent financial institutions to adopt flexible technology, and inspiring non-financial companies to begin offering credit through partnerships. A plethora of tools is now available for different dimensions of digital transformation, but navigating them can be difficult. Technology can be expensive and hard to vet, and many financial institutions worry about making the wrong choice, or even adopting the right tools in the wrong order. 

Credit scoring can be overwhelming for an institution that is just beginning to digitize. A lot of big data in Africa is not yet structured in a way that is complete, consistent, and easily accessible. Drawing on data from telcos, credit bureaus and other sources requires developing integration platforms with multiple partners, which can be expensive and bureaucratic. Once all that data is accessible, it must also be processed and analysed to meet the specific needs of the loan provider, which requires expertise in analytics and the software tools to design and run models.

First Access learned a few major lessons in its R&D and work with partners: 

  • Digitize from the start: Financial institutions should digitize 100% of the data they record about customers immediately, instead of writing it on paper and keying 5-10% of it into their MIS after loan approval. This data is not only an important asset because it’s predictive and provides real-time operational and market insights, but also because it is already trusted by staff, free, and immediately available. 
  • Make the most of internal data: For loan products that rely on human judgment, as most agricultural loans do, this ‘internal’ data should be used to improve credit decisions before drawing on external data. It’s the best way to make early gains in the digital journey, like fast-tracking repeat customers with no late days. Lenders have used First Access to save customers over 300 years of wait time through fast-tracking. 
  • Build a data culture: As staff become familiar with collecting data digitally, data-driven decision making like fast-tracking loans, and other such processes, they also build a data culture. This allows them to adopt increasingly sophisticated tools, data, and processes with less training. It also makes a lender more likely to maximize the benefits of its own data before investing in more expensive, riskier external sources of data that generally require more internal know-how.

Internal data can be easily collected on the First Access platform, which, with support from FRP, can now be fully customized and deployed for agricultural loans in 2 to 4 weeks. It allows lenders to draw effectively on their own data to make credit decisions, and when they are ready, to invest in new products, data sources, and partnerships to reach more customers.

First Access is now working with agricultural lenders in 3 countries, having migrated their paper-based processes quickly and easily to a flexible mobile app and web platform. Setup costs are relatively low, and the service is maintained over time through a multi-year fee or with financing through a monthly subscription agreement. For lenders that are accustomed to buying expensive custom software like core banking or DFA (dynamic financial analysis) systems, this model of purchasing software requires a shift in mindset and a willingness to try a low-cost pilot. Implementation also requires less change management, since the platform is first mapped to existing processes for easy staff training, then evolved over time according to business needs.

The First Access platform allows loan officers to collect information easily onsite at the borrower’s home, business, farm, and other locations, even offline, including geolocation tags, photos, and crop data. It can also be used to retrieve information from external sources, like agricultural processors or input providers with transactional data on smallholders. Staff can submit loans for instant review and approval by designated managers. This reduces loan turnaround time, and some lenders can already approve a low-risk customer within a day instead of a week.

The platform can include automated risk assessments or scoring, which is set up according to an institution’s risk appetite and conditions, and can route customers to different levels of loan diligence and approval requirements. This allows the financial institution to scale faster by allocating time and resources based on customer segments. Over time, lenders can leverage their data to improve portfolio quality, efficiency, speed, and the number of customers per loan officer.

This rules-driven, digitized workflow approach can automate some or all credit decisions, and it doesn’t necessarily require credit scoring, which indexes each person against the population, requires data that is greater in quantity and quality, and is hard to use effectively without a data culture. This predictive modelling is also a highly consultative and expensive approach, which First Access has found provides a level of analysis that many financial institutions simply don’t need yet, and which they want to drive themselves when they’re ready. 

Of course, the market is evolving. As more data sources and APIs (application programming interfaces, which facilitate access to data) become available and more financial institutions get comfortable with digital processes, they can use First Access to import and analyse external data, such as transaction records from a pool of new potential customers, who can then be pre-qualified or even automatically approved. These new data sources will make credit more accessible and affordable, and bring a substantial competitive edge to lenders who have adopted flexible technology and built a data culture.

Case Study written by Rachel Keeler, Project Manager, Mastercard Foundation Fund for Rural Prosperity with contributions from Nicole Van Der Tuin, Co-Founder and CEO, First Access and Corey Pargee, Director of Operations, First Access

Add new comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.