Data is ubiquitous. Especially in agriculture.
But data with economic value attached? Not so common.
Think of all the big data that’s out there - 72,000 searches are processed per second by Google (or 6.2 billion per day); 77,000 videos are played any given second of the day on Youtube. This gives off the big data - the data that feeds Google algorithms and keeps their enormous advertising business running.
Agriculture needs small data. In an academic paper titled “Big Data in Agriculture: A Challenge for the Future”, the authors introduce the concept of small data. This small data is in-field agricultural variants (soil sample data, yield data, drone data, weather data). The problem is, it becomes “isolated to the fields where the data originated”. Aggregating this data is pivotal to reassuring banks that lending to agriculture and allied activities is probably not as risky.
Banks have historically held agricultural lending precariously.
And despite the sector being the top beneficiary in the Priority Sector Lending (PSL) mandate for banks by the Reserve bank of India (RBI), banks face challenges in meeting their requirement due to
High perceived risk of default
Non-digitisation of farmlands
Increased cost of verification
Unavailable credit history
Collections are tricky
Loans are used for non-agricultural purposes such as education, healthcare expenses and marriage
And the data proves how formal/informal sources are jittery to lend to small and marginal farmers -
Only 40% of India's small and marginal farmers are covered by formal credit
Around 45% of all Indian farmers possess an operative Kisan Credit Card (KCC).
50% of India’s small and marginal farmers are unable to borrow from any source — tech or traditional
Of the $168 Bn agriculture credit offered by banks in FY19, over half was offered to medium and large farmers, who already have access to formal capital.
These Small and marginal farmers make up for 86.2% of the farmer economy.
Small and marginal farmers are spread out across large, thinly populated areas, making them costly to reach and serve; they have few assets to pledge as collateral against loans and with no financial history, they’re deemed risky by banks. And on top of all that, they’re often at the mercy of the weather.
Yet, farmers with access to credit can invest in fertilizers and seeds that could increase their yields and incomes by more than 25%.
Fintechs have been running on alternate credit scoring systems to bring new-to-credit customers into the financial system. What data can agriculture generate to help banks lend without sweating?
Government survey numbers have digitized plots in select states of India. This can be used to verify the existence and location of the plot and lenders can ensure that the plot is used for agriculture using AI algorithms.
Crop suitability mapping
Location information is critical to agriculture. In February 2021,the Indian government opened access to its geospatial data and mapping services for all Indian entities. Machine learning algorithms help in crop scouting, soil sampling, weed location, accurate planting and harvesting. using geo-mapping techniques, lenders are able to ensure that they only lend to a farmer growing a suitable crop in a suitable location.
Weather is the biggest risk in agriculture. Historical weather and temperature range analysis for a given plot is an important data point for underwriting.
Real time monitoring analytics
Banks need to track loans throughout the crop season on multiple parameters - growth stage, exact sowing data, expected harvest date. The data that comprises these parameters are captured by smart sensors and drones that provide real-time video streaming that gives lenders access to data sets they’ve never had access to before. Machine learning, combined with AI, in-ground sensors and infrared imagery is the perfect to combine massive data sets and provide real-time insight.
UPI for feature phones
UPI for feature phones is another key event in solidifying India’s financial inclusion plan. And no financial inclusion plan is complete without accounting for the agrarian economy. The true last mile in that economy is the small and marginal farmers' economy. For small and marginal farmers, no access to credit comes in the form of untenable visits to a bank branch, lack of financial literacy (even as basic as using an ATM), no access to smartphones or the internet. The fintech innovation that will transpire on top of the new UPI features (voice response (IVR) numbers, feature phone apps, missed calls, and sound-based payments) could help lenders underwrite better.
These data points can fit into a farm analysis package to generate an alternate credit scoring system - a field score of sorts. This field score can help lenders make data-backed, informed credit underwriting decisions. This way lenders can ensure loans are issued to productive farmers (who may not get loans because of credit history) and help banks meet their PSL requirements as well.
Fintechs that want to underwrite for lenders can either deploy AI, ML, IOT technologies to develop this field score via partnerships with Farmer Producer Organizations (FPOs) or they can do so via partnerships with several up and coming ag-tech companies.
Either way, data prudence is the backbone of agricultural financing.
And fintechs are evolving capabilities to develop complex economic models that take into account how monsoons will behave, irrigation density, reservoir levels, soil moisture levels, MSPs, what the state’s finances look like, alternative sources of income available to farmers, etc.
We think data-driven agricultural financing will be a sector of opportunity for years to come. It’s really only a matter of intent and a favorable policy environment.