A Kenyan woman holds a phone displaying the Apollo Agriculture app
Satisfied customer: Apollo Agriculture uses AI to evaluate borrowers’ ability to repay © Apollo Agriculture/Kenyafixer

While they produce about a third of what we eat, the world’s smallholder farmers have long struggled to secure loans. That, though, is changing. Thanks to a “boots-on-the ground” approach and the use of technologies such as artificial intelligence and machine learning, new lending models are being developed that can expand rural access to finance.

For traditional lenders, extending credit to smallholders is not easy. They must contend with variables ranging from rising central bank interest rates to weather-related risks and the high cost of administering small-scale loans.

“When we got started, one of the things we noted was that even well-intentioned microfinance banks weren’t lending at any volume to small-scale farmers,” says Eli Pollak, chief executive of Apollo Agriculture, a Kenya-based agritech company. “Fundamentally, their cost structures are suited for a much larger customer.”

However, Apollo can assess farmers’ ability to repay loans by applying AI machine learning to data sources that include satellite images, third-party credit ratings, and information gathered through a 5,000-strong network of agents working on commission.

It then provides loans in the form of vouchers that can be spent on farm inputs — such as seeds and fertilisers.

Advances in observation data have contributed significantly to these new models of agricultural micro-lending. “Historically, to see what was happening on a farm, you had to go there,” says Pollak. “Now, machine learning tools allow us to generate real insights from satellite data.”

Next-generation satellites can generate high-resolution images of any part of the Earth in detail, often in near real time. And these can help lenders to make better assessments of one of the biggest risks to farmers: the vagaries of the weather.

For example, Norwegian software company Sensonomic is working with non-profit microfinance institution Opportunity International and the European Space Agency to develop a commercial service for agricultural lenders that harnesses data capture, data analytics, and advanced simulation, to improve assessments of future yields.

Envisat image showing a green rectangle of cultivated land among volcanoes in eastern Africa
How’s it growing?: an Envisat image shows cultivated land near volcanoes in eastern Africa © ESA

When it comes to traditional micro-lending, technology can help overcome another barrier: the lengthy approvals process that can make or break small-scale farms.

In Uganda, Opportunity International agents equipped with iPads can access credit scorecards and behavioural analytics to run credit checks while on the farm. This has cut loan approval times from 60 days to four, says Timothy Strong, Opportunity’s global head of agricultural finance.

An additional roadblock to agricultural microfinance has been the requirement for collateral — land, vehicles or buildings — as the basis for a loan. Smallholder farmers rarely have — or are able to prove — full ownership of such assets.

But, in Ghana and Kenya, loan origination and credit app Mfarmpay is addressing the problem by applying machine learning to satellite images, climate metrics and information that farmers provide on their crops and the extent of their land.

Using this data to assess farmers’ risk of crop failure and the climate resilience of their agricultural practices, Mfarmpay generates a credit score that smallholders can use as an alternative to collateral.

Loans are not the only facilities needed to raise farmers’ incomes, though.

“You need to take a holistic approach,” argues Nicole Van Der Tuin, chief analytics officer at the microfinance group Accion Opportunity Fund. “Just solving for better pricing of loans for lenders doesn’t seem to take off as an independent business model,” she says.

For example, as well as extending loans to farmers, Apollo provides seeds and fertilisers, insurance products and advisory services.

Similarly, at Mfarmpay, the bundled model includes digital advisory services and market information. “For every loan, they’re guaranteed off-taking of the commodity,” explains co-founder and chief executive Elorm Allavi.

This in turn makes it easier for farmers to access credit. “It de-risks the loan facility and further incentivises banks,” says Allavi.

In India, impact-driven agritech company Harvesting Farmer Network is using a combination of technology and on-the-ground knowledge. To scale up use of the data-driven intelligence that is needed to expand access to credit, HFN is creating a next-generation agricultural co-operative.

Unlike their traditional counterparts, these HFN co-operatives are not necessarily physically proximate but can be connected digitally. As a result, the more than 3.7mn farmers in the network have the same strength in numbers: increased their purchasing power and their negotiating clout when selling crops.

HFN marries its first-hand understanding of farmer economics with technologies such as remote sensing, AI and mobile data to create what it calls its “intelligence engine” — a tool that the banks it partners with can use to expand their micro-lending.

“Our data enables the banks to better service these farmers,” says Ruchit Garg, the company’s founder and chief executive. “We’re focused on making information more accurate, transparent, and cheaper or free. If you can make that information easily available, you can democratise access to finance.”

Copyright The Financial Times Limited 2024. All rights reserved.
Reuse this content (opens in new window) CommentsJump to comments section

Follow the topics in this article