Bridging the data gap in agricultural financing: The case of FarmDrive

FarmDrive is a data analytics and Fintech company, founded in Kenya in 2014, that supports small-scale farmers in accessing credit from local financial institutions (FIs). The aim of the company is to bridge the gap that traditionally separates small-scale agricultural producers from the formal financial sector, which, among other factors, is caused by the scarce data and information that financial institutions have on these actors – which makes them unable to carry out accurate credit profiling for small-scale farmers and agri-enterprises. As a result of this information asymmetry, most farmers in developing countries are prevented from obtaining loans from formal FIs – even when they would be perfectly capable of repaying one.
FarmDrive works through a mobile platform that aggregates and analyses information sourced from both traditional and alternative data points, using machine learning[1] to generate accurate credit profiles of farmers and their capacity to repay, even when they lack an established history of borrowing through the formal financial sector. Through the platform, FarmDrive acts as an intermediary and facilitator between a network of partner FIs and a large client base of farmers applying for credit. Thanks to FarmDrive’s credit profiling and intermediation services, formal FIs can expand their agricultural loan portfolios while mitigating the associated credit risk, as well as reduce the overall operational costs associated to lending to small-scale agriculture. Farmers, on the other hand, are able to access high-quality loans from the formal financial sector for the first time, through the company’s mobile platform.
At the beginning of the credit profiling process, FarmDrive collects farm-level data relative to the loan applicant from a variety of different sources, which is then combined and processed by yield-prediction algorithms that have been designed on the basis of the specific agronomic features of the different areas where FarmDrive’s clients reside, as well as the crops they cultivate. More specifically, the data collected at this stage is categorized according to three streams:
- Agronomic data: which includes extent and variety of crops cultivated at farm level, soil quality, drainage infrastructure, weeds, pests, etc.
- Remote-sensing data: which provides information on vegetation quality, weather patterns, temperature, etc.
- Market data: which includes price trends, offtake security, etc.
The results of the yield-prediction algorithm are then combined with behavioral data that integrates information on the habits, perspectives, and attitude towards repayment of the farmer applying for the loan. This data is collected from farmers through a mobile service, accessible either through a basic handset (via SMS or USSD[2]) or a smartphone app. In fact, the entirety of FarmDrive’s platform services can be accessed through a basic mobile handset, which represents an important factor for the company’s outreach in rural areas. The combination of farm-level and behavioral data allows FarmDrive to develop highly accurate credit profiles of potential borrowers engaged in small-scale agriculture, while assisting the partner financial institutions in making the most accurate decisions when selecting the parameters of a specific loans (i.e. loan terms, period, amount, grace period, etc.). Furthermore, farmer clients are able to log and track their business’ productivity, expenses and revenues through the platforms. With the farmer’s consensus, this information can be access and analyzed by the partner financial institutions – to highlight performance patterns and assist them in refining the loan approval process.
The revenue model of FarmDrive is based on two main sources: a fixed fee paid by the partner FIs when they make use of the company’s credit profiling services, and a transaction fee paid by the client farmer when he receives a loan through the platform – expressed as a percentage of the total loan amount intermediated by FarmDrive. It must be noted that FarmDrive’s capital expenses are very low, as its model is mostly digital. The main operating expenses in its business model are associated to training the farmers on the use of the mobile application, as well as the cost of creating the credit profiles.
At the end of 2020, FarmDrive’s services had unlocked over USD 10 million in loans for more than 100 000 Kenyan smallholders and agri-enterprises. Furthermore, the company had created a digital transaction registry of more than 1.2 million farmers that formal financial institutions could peruse, in order to find potential borrowers and expand their agricultural credit portfolios. The company’s growth in recent years has also been supported by substantial investment capital raised by impact investment funds and similar entities. In 2019, for example, FarmDrive received USD 3 million raised through an investment round led by EWB (Engineer Without Borders) Canada, with the participation of AK Impact Investors and ADAP Seed Fund 2. In 2017, it received an undisclosed amount through an investment made by SafariCom Spark Venture Fund.
FarmDrive is yet another interesting example of a rapidly growing number of Fintech companies that employ a combination of machine learning, alternative data sources and mobile technology to bridge the gap between the formal financial sector and developing agriculture, in an attempt to capture the enormous market potential associated to the unmet demand for credit on the part of small-scale agricultural actors (a USD 74 billion financing gap in Sub-Saharan Africa only, according to the latest estimates from ISF Advisors).
[1] Machine learning is a method of data analysis that automates analytical model building. It is a branch of artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.
[2] USSD (Unstructured Supplementary Service Data) is a Global System for Mobile Communications (GSM) protocol that is used to send text messages, similar to SMS. USSD uses codes made up of the characters that are available on a basic mobile handset. A USSD message, which can be up to 182 characters long, establishes a real-time communication session between the phone and another device — typically, a network or server. With USSD, users interact directly from their mobile phones by making selections from various menus. Unlike an SMS message, during a USSD session, a USSD message creates a real-time connection. This means that USSD enables two-way communication of information, as long as the communication line stays open. As such, queries and answers are nearly instantaneous.