Table of contents
Agriculture is the sum of many parts. Its success depends on a range of factors from soil quality and weather vagaries to commodity markets and policy interventions. This is precisely what makes agriculture a seemingly risky sector for lenders – there’s no telling which of these factors may blindside farmers, and by extension, their creditors.
However, the diversity of factors that determine agricultural success or failure can be a potent tool in the hands of digital lenders. But the problem is, these data sets are as fragmented as they are diverse. And, so far they have existed in silos unbeknownst to one another.
Mobile apps, devices, systems and other repositories of these data sets must talk to each other. Possible outcomes of interoperability among these datasets include the realization of the GoI’s ambitious plan to double farmer incomes, banks’ priority sector lending imperatives and the vision of financial inclusion.
Components of a field score
Unconventional data is at the heart of digital lending. Sectors like agriculture, particularly, that have historically been starved of credit due to lack of adequate data for underwriting stand to benefit from alternate data-based digital lending. However, the datasets required for agricultural lending are wildly disparate.
Let’s take a look at some of the types of data required to score farmers –
Land records: Any records relating to land ownership, rights, disputes or other concerns like the record of rights, the register of land, disputed cases register or tenancy records. There is a country-wide drive to digitize these records, with states like Karnataka, Rajasthan, Kerala UP, Gujarat, Madhya Pradesh, Andhra Pradesh and Telangana having digitized their land records completely. This would facilitate faster direct disbursement of loans to farmers.
Transaction records: Records of produce bought and sold at various points in the supply chain can now be culled digitally. These include records of the nature and quantity of crops purchased by traders and the agricultural inputs sold to farmers. This data can be sourced from buyers’ ledgers, digital payment portals or e-wallets.
Survey data: Data points like the income of agricultural households, landholdings and livestock ownership, farming practices, demographics and land usage can be accessed from government surveys conducted by the National Sample Survey or the agricultural census.
Demographic data: KYC requirements such as date of birth, gender, marital status, etc. can shed light on the creditworthiness of various farmer demographics. Although such data often feeds into lending biases, by re-engineering machine learning models, it can lend useful insights into creditworthiness.
Satellite imagery and geospatial data: Data around soil quality, irrigation facilities, climate, plot size and crop type can be gathered from government resources like Bharat Maps, meteorological companies like The Weather Channel or even through partnerships with new space-tech startups.
Alternate data: Big data including metadata, social media activity, and other intelligence sourced from mobile phone activity like the nature of downloaded apps, income estimates and such has proved effective in underwriting for other priority sectors like SMEs. It has similar applications in agri underwriting.
Credit scores: Credit history including the number of open loan accounts, credit utilization, quality and repayment. All these factors make up traditional credit scores that can be sourced from credit bureaus.
Some opportunities in agri lending
Getting different systems in possession of these varied data to communicate can unlock a number of use cases in agri lending. Here are a few opportunities arising out of interoperable systems.
Access to various datasets can reduce information asymmetry in the supply chain, empowering each player from farmer to end consumer. Such enfranchised supply chain players are better positioned to seek out financing that can allow them to take bigger risks and reap the rewards.
Interoperability opens up access to several unconventional data sources for underwriting, creating an opportunity to score new-to-credit farmers.
Communication among various systems can make digital agri lending more agile. For instance, by partnering with various data stewards a lender can draw up an alternate credit score within seconds and prequalify eligible customers within the app.
Interoperability among platforms and lenders can allow underwriting based on platform data and offer proprietary financial solutions to the platform’s customers.
The challenge facing agri fintech companies is how to unite this overwhelmingly variegated data and draw insights relevant to lending from it. The systems and software holding this data need to be made compatible technically (physical infrastructure for data sharing), syntactically (similar structure and format of data storage) as well as semantically (ease of understanding the data; shared meaning).
These are being addressed with the use of open source software that allows for peer review, community production and collaboration. The development of a standardized language for exchange of agri data such as agroXML by KTBL in Germany is another example of agricultural systems being able to communicate. Closer home, the proposed Agristack will link public data from existing schemes with digitized land records.
A common language for data sharing in agriculture has many benefits for agriculture, foremost among them financing. Interoperability can make digital agri lending agile (eg. fast underwriting based on various datasets for in-app prequalification) and democratize finance (eg. credit scoring for new-to-credit farmers based on unconventional data). The time to innovate is truly here and arguably, a lot more innovation is needed in getting data to speak to each other than collecting even newer sources.